Vu Thesis

2012-ENST-00xx
EDITE - ED 130
Doctorat ParisTech THSE

pour obtenir le grade de docteur dlivr par
TELECOM ParisTech
Spcialit Informatiques et Rseaux
prsente et soutenue publiquement par
Thanh Tung VU
20/09/2012
Modles spatiaux pour la planication cellulaire
Directeurs de thse: Laurent DECREUSEFOND Philippe MARTINS
Jury Mme/M. Mme/M. Mme/M. Mme/M. Mme/M. Mme/M. Mme/M. Mme/M.
Prnom Prnom Prnom Prnom Prnom Prnom Prnom Prnom
NOM, NOM, NOM, NOM, NOM, NOM, NOM, NOM,
Titre, Titre, Titre, Titre, Titre, Titre, Titre, Titre,
Unit Unit Unit Unit Unit Unit Unit Unit
de de de de de de de de
recherche, recherche, recherche, recherche, recherche, recherche, recherche, recherche,
Ecole Ecole Ecole Ecole Ecole Ecole Ecole Ecole
Fonction Fonction Fonction Fonction Fonction Fonction Fonction Fonction
TELECOM ParisTech
cole de lInstitut Tlcom - membre de ParisTech
Spatial models for cellular network planning

Thanh Tung Vu
Rsum
Aujourdhui, la technologie cellulaire est peu prs partout. Il a eu un succs explosif au cours des deux dernires dcennies et le volume du trac va encore augmenter dans un proche avenir. Pour cette raison, il est galement considr comme une des causes de la consommation dnergie dans le monde entier, avec un fort impact sur les missions de dioxyde de carbone. Dautre part, de nouveaux outils mathmatiques ont permis construire de nouveaux modles pour les rseaux cellulaires : un de ces outils est la gomtrie alatoire, ou plus particulirement processus spatiale de Poisson. Dans la dernire dcennie, les chercheurs ont utilis avec succs la gomtrie alatoire pour quantier probabilit doutage, le dbit ou la couverture des rseaux cellulaires en traitant le dploiement de stations mobiles ou stations de base (et) en tant que processus ponctuels de Poisson sur un plan. Ces rsultats prennent galement en compte limpact de la mobilit sur la performance de ces rseaux. Dans cette thse, nous appliquons la thorie de processus de Poisson pour rsoudre certains problmes de rseaux cellulaires, en particulier, nous analysons la consommation dnergie des rseaux cellulaires. Cette thse comporte deux parties principales. La premire partie est consacre rsoudre quelques problmes de dimensionnement et de couverture des rseaux cllulaires. Nous calculons la probabilit de surcharge des systmes OFDMA grace aux ingalits de concentration et lexpansion dEdgeworth et nous lappliquons rsoudre un problme de dimensionnement. Nous calculons galement la probabilit doutage et la taux dhandover pour un utilisateur typique. La seconde partie est consacre prsenter des modles dirents pour la consommation dnergie des rseaux cellulaires. Dans le premier modle, lemplacement initial des utilisateurs forme un PPP et chaque utilisateur est associ un processus de marche-arrt (ON-OFF) de lactivit. Dans le second modle, larrive des utilisateurs constitue un espace-temps processus de Poisson. Nous tudions galement limpact de la mobilit des utilisateurs en supposant que les utilisateurs se dplacent de manire alatoire pendant ses sjour. Nous nous intressons la distribution de lnergie consomme par une station de base. Cette nergie est divis en deux parties : la partie additive et la partie disfuse. Nous obtenons des expressions analytiques pour les moments de la partie additive ainsi que la moyenne et la variance de lnergie totale consomme. Nous trouvons une borne derreur pour approximation gaussienne de la partie additive. Nous prouvons que la mobilit des utilisateurs a limpact positif sur la consommation dnergie. Il naugmente pas ni rduit lnergie consomme en moyenne, mais rduit sa variance 0 en rgime de mobilit lev. Nous caractrisons aussi le taux de convergence en fonction de la vitesse des utilisateurs.
Abstract
Nowadays, cellular technology is almost everywhere. It has had an explosive success over the last two decades and the volume of trac will still increase in the near future. For this reason, it is also regarded as one cause of worldwide energy consumption, with high impact on carbon dioxide emission. On the other hand, new mathematical tools have enabled the conception of new models for cellular networks: one of these tools is stochastic geometry, or more particularly spatial Poisson point process. In the last decade, researchers have successfully used stochastic geometry to quantify outage probability, throughput or coverage of cellular networks by treating deployment of mobile stations or (and) base stations as Poisson point processes on a plane. These results also take into account to impact of mobility on the performance of such networks. In this thesis, we apply the theory of Poisson point process to solve some problems of cellular networks, in particular we analyze the energy consumption of cellular networks. This thesis has two main parts. The rst part deals with some dimensioning and coverage problems in cellular network. We uses stochastic analysis to provide bounds for the overload probability of OFDMA systems thanks to concentration inequalities and we apply it to solve a dimensioning problem. We also compute the outage probability and handover probability of a typical user. The second part is dedicated to introduce dierent models for energy consumption of cellular networks. In the rst model, the initial location of users form a Poisson point process and each user is associated with an ON-OFF process of activity. In the second model, arrival of users forms a time-space Poisson point process. We also study the impact of mobility of users by assuming that users randomly move during its sojourn. We focus on the distribution of consumed energy by a base station. This consumed energy is divided into the additive part and the broadcast part. We obtain analytical expressions for the moments of the additive part as well as the mean and variance of the consumed energy. We are able to nd an error bound for Gaussian approximation of the additive part. We prove that the mobility of users has a positive impact on the energy consumption. It does not increase or decrease the consumed energy in average but reduces its variance to zero in high mobility regime. We also characterize the convergent rate in function of users speed.
Contents
Page 1 Introduction 1.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.2 Thesis outline and contributions . . . . . . . . . . . . . . . . . . . . . . . . . 1.3 Notations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 11 12 14
Mathematical background and contributions
17
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . expansion . . . . . . . 28 . . . . . . . 31 . . . . . . . 33 19 19 19 21 23 26 26
2 Poisson point process 2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2 The use of Poisson point process on wireless modelling . . . . . 2.3 Poisson point process . . . . . . . . . . . . . . . . . . . . . . . . 2.4 Malliavin calculus on Poisson point process . . . . . . . . . . . 2.5 Distribution of linear functional of Poisson point process . . . . 2.5.1 Moments . . . . . . . . . . . . . . . . . . . . . . . . . . 2.5.2 Existing results on Gaussian approximation, Edgeworth and concentration inequality . . . . . . . . . . . . . . . . 2.5.3 New results . . . . . . . . . . . . . . . . . . . . . . . . . 2.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
II
Dimensioning and coverage models

dimensioning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
35
37 37 39 42 42 42 44 44
3 Robust methods for LTE and WiMAX 3.1 Introduction . . . . . . . . . . . . . . . 3.2 System Model . . . . . . . . . . . . . . 3.3 Loss probability . . . . . . . . . . . . . 3.3.1 Exact method . . . . . . . . . . 3.3.2 Approximations . . . . . . . . . 3.3.3 Robust upper-bound . . . . . . 3.4 Applications to OFDMA and LTE . .
4 An analytic model for evaluating outage and handover cellular wireless networks 4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2 System model . . . . . . . . . . . . . . . . . . . . . . . . . 4.3 Outage analysis . . . . . . . . . . . . . . . . . . . . . . . . .
probability of 49 . . . . . . . . . 49 . . . . . . . . . 50 . . . . . . . . . 52
8 4.4 4.5 4.6 5 On 5.1 5.2 5.3 5.4 5.5
CONTENTS Handover analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54 Numerical results and comparison to the hexagonal model . . . . . . . . . . 55 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59 59 60 61 65 70 70 71 71 71 72 73
noise-limited networks Introduction . . . . . . . . . . . . . . . . . . . Model . . . . . . . . . . . . . . . . . . . . . . Poisson point process of path loss fading . . . Capacity . . . . . . . . . . . . . . . . . . . . . Examples . . . . . . . . . . . . . . . . . . . . 5.5.1 Number of users in a cell . . . . . . . . 5.5.2 Number of users in outage in a cell . . 5.5.3 Number of covered users in a cell . . . 5.5.4 Total bit rate of a cell . . . . . . . . . 5.5.5 Discussion on the distribution of So (f ) 5.6 Conclusion . . . . . . . . . . . . . . . . . . . .
III
Energy consumption models

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
75
77 77 77 80 80 84 85 85 86 88 93 98 98 99 102
6 Generality and basic model 6.1 Introduction . . . . . . . . . . . 6.2 Model for energy consumption . 6.3 Model for mobility of users . . 6.4 Basic model . . . . . . . . . . . 6.5 Conclusion . . . . . . . . . . . . 7 ON-OFF model 7.1 Introduction . . . . . . . . 7.2 Model . . . . . . . . . . . 7.3 Motionless case . . . . . . 7.4 Impact of mobility . . . . 7.5 Special cases . . . . . . . 7.5.1 Completely aimless 7.5.2 Always on users . . 7.6 Summary and Conclusion
. . . . . . . . . . . . . . . . . . . . . . . . . mobility . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . model . . . . . . . .
8 Generalized Glauber model 8.1 Introduction . . . . . . . . . . . . . . . . . . . . . . 8.2 Model description and main results . . . . . . . . . 8.2.1 Generalized Glauber dynamic . . . . . . . . 8.2.2 Generalized Glauber dynamic with mobility 8.3 Analysis in no mobility case . . . . . . . . . . . . . 8.3.1 Proof of Theorem 48 . . . . . . . . . . . . . 8.3.2 Proof of theorem 49 . . . . . . . . . . . . . 8.3.3 Proof of theorem 50 . . . . . . . . . . . . . 8.4 Impact of mobility . . . . . . . . . . . . . . . . . . 8.4.1 Lemmas . . . . . . . . . . . . . . . . . . . . 8.4.2 Proof of Theorem 53 . . . . . . . . . . . . .
105 . 105 . 106 . 106 . 109 . 110 . 115 . 116 . 116 . 116 . 116 . 119
9 8.4.3 Special case: Completely aimless mobility model . . . . . . . . . . . 120 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
8.5
IV
Conclusion
123
9 Conclusion and future works 125 9.1 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125 9.2 Future works . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126 10 Appendix 129 10.1 ON-OFF exponential process: Basic properties . . . . . . . . . . . . . . . . 129 10.2 Hermite polynomials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131 10.3 Ornstein-Uhlenbeck semi-group . . . . . . . . . . . . . . . . . . . . . . . . . 131 Bibliography 141
10
CONTENTS
11
Chapter 1
Introduction
Contents
1.1 1.2 1.3 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 Thesis outline and contributions . . . . . . . . . . . . . . . . . . 12 Notations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
1.1
Motivation
Cellular communications have realized an amazing evolution for the last twenty years. Technological advances in cellular systems and cellular phones design have made possible services that one could not imagine twenty years ago. Mobile telephony, and more particularly GSM, have been the rst worldwide service with more than 5 billions customers today. The third generation systems such as UMTS have tried to provide a universal radio interface both adapted to circuit and data services transfer. Success of 3G technologies have been quite disappointing at the beginning. However, recent developments such as HSPA, HSPA+ and fourth generation systems tend to show that cellular communications are now at the eve of a new major revolution. These new technologies will change the way of life of many people by bringing them new convenient utilities and services. Modern smartphones are now built on mobile computing platforms, with more advanced computing and connectivity abilities than ever and at aordable prices. People are now able to use data services such as localization, Internet access, les transferring from almost anywhere even in high mobility. Data trac associated with these services is more and more important. From a network operator point of view resources are necessary to satisfy user demands and fulll quality of service requirements. Such an evolution requires more and more spectrum resources but unfortunately this resource is scarce. As a result network operators need to design more and more ecient methods to design and deploy their network. Cellular planning is the process used to design, dimension and deploy a mobile network taking into quality of service constraints. This is a complex procedure that requires to take into account many criteria : An operator needs to satisfy trac and QoS demands of its customers. To fulll that objective, it is necessary to deploy resources such base stations, antennas, routers, switches or transmission links. Dimensioning is the process that determines the number of resources necessary to satisfy trac and QoS constraints.
12
1. Introduction Coverage is the process that grants that a particular cell provides the necessary radio quality to comply with users QoS constraints. Capacity determines the number of subscribers that can be supported by a network taking into account QoS and radio constraints.
This work intends to provide new models that can be applied in the cellular planning process. The proposed solutions model and take into account the eect of spatial distribution of users and bases stations. The investigations carried out in this work rely signicantly on stochastic geometry, point process theory and probability distribution approximation methods. The most important contributions of these investigations can be classied into the following three areas: This thesis proposes new spatial models for the performance evaluation of several aspects of OFDMA networks. The rst problem addressed is the dimensioning of an OFDMA cell in terms of resource blocks or subchannels. The purpose is to nd analytical expressions of the blocking probability. Two dierent approaches have been considered. In a rst approach, upper bounds of blocking probability have been computed. These bounds are quite coarse but always overdimension the resources need in the cell. That excess of resources has an advantage. It provides some robustness against the inaccuracy of radio propagation modeling. A second approach relies on approximations methods (Egdeworth expansions). They give accurate estimations of the blocking probability for well known radio propagation parameters but at the expense of a lack of robustness. Outage and handover are both critical issues in cellular systems. Outage is tightly related with the coverage issue. Outage probability provides an indicator of the coverage quality in a network. This thesis develops outage probabilities expressions more general than classical expressions obtained in hexagonal or Voronoi Tessellations models. The performance of handover is also a central issue. That metric is necessary to have an estimate of the handover trac and hence to dimension cells properly. Closed forms for handover probability are also developed in this investigation work. Finally energy consumption in cellular networks is also considered in that work. The thesis proposes models that make it possible to estimate energy consumption in a cell taking into account trac and spatial distribution of users. The proposed models can be used to dimension sites that do not have access to power supply facilities.
1.2
Thesis outline and contributions
This thesis consists of two parts. The rst part introduces the elements of Poisson point process theory and applications to the dimensioning the OFDMA systems. The second part applies the theory presented in the rst part to study the energy consumption of cellular networks. Chapter 2 gives an introduction and results on Poisson point process theory that we use throughout this thesis. We rst dene Poisson point process in an understandable way and present some of its important properties such as the distribution of the number of points and the Campbell Theorem or some operations preserving the Poisson law. We then study the distribution of linear functionals depending on a Poisson point process as many
13 functionals of interest have this form. We present an upper bound on tail distribution (called concentration inequality) and an error bound of Gaussian approximation of such functionals. We then turn to the Malliavin calculus on Poisson point process. It allows one to decompose a large family of functional depending on Poisson point process as the sum of orthogonal chaos, where the nth chaos is the contribution of every set of n points of the process. Bounds on the tail distribution and Gaussian approximation of a general functional is also presented, that generalize bounds in the linear case. The chapter 3 proposes an analytic model for dimensioning OFDMA based networks like WiMAX and LTE systems. In such a system, users require a number of subchannels which depends on their SNR, hence of their position and the shadowing they experience. The system is overloaded when the number of required subchannels is greater than the number of available subchannels. We give an exact though not closed expression of the loss probability and then give an algorithmic method to derive the number of subchannels which guarantees a loss probability less than a given threshold. We show that Gaussian approximation leads to optimistic values and are thus unusable. We then introduce Edgeworth expansions with error bounds and show that by choosing the right order of the expansion, one can have an approximate dimensioning value easy to compute and with guaranteed performance. As the values obtained are highly dependent from the parameters of the system, which turned to be rather undetermined, we provide a procedure based on concentration inequality for Poisson functionals, which yields to conservative dimensioning. This chapter relies on recent results on concentration inequalities and establish new results on Edgeworth expansions presented in the Chapter 2. Chapter 6 presents a general energy consumption model for a base station in a cellular network. We rst introduce the power consumption as a function of the conguration of active users at each time. The energy consumed during a time period is then dened as the integral of the consumed power during this duration. We divide the consumed energy into two parts: the broadcast part served to transmit the same message to all users in the cell, and the additive part which sums up all the energy used by each user both in downlink and uplink modes. We then introduce the general model for users mobility, where trajectories of users are i.i.d. Finally we study a basic model where for each instance the conguration of active users is a Poisson point process. We show that the consumed energy is an increasing function of the cell radius and that there is always an optimal cell radius in the economical point of view. We then study the energy consumption of cellular network in two models for dynamic of users. In the chapter 7, we assume that each user is associated with an activity on-o process in time. In the chapter 8 we no longer assume that a user is on or o. Rather we assume that users arrive following a Poisson point process in space and time and make communicationq during certain time being modeled by a random variable. We are able to provide analytic expressions for statistics of consumed energy and bounds on its distribution. We then consider the impact of mobility and we show that when users move, the average consumed energy does not change while its variance decreases to zero when users move fast. This is a strong result as it holds true for any mobility model. In the on-o model we are able to characterize the decay rate of variance in function of users speed. In both models, we provide asymptotic expressions when the system works for a very long time.
14
1. Introduction
1.3
Notations
Tables 1.1 and 1.2 present all mathematical notations and abbreviations used throughout this thesis.
15
Symbols R+ D P(A) E [X ] V [X ] mn [X ] cn [X ] pX (t) FX (t) F X (t) X n B(x, r ), Bd (x, r ) B(x, r ), Bd (x, r ) 1{A} (x) C [X, Y ] o C C (r ) C (r ) dW (X, Y ) Q(u) = Q(u) =
1 2 1 2 u x 2 e x 2 u e
2 2
Denition [0, ) dierential operator probability of event A expectation of random variable X variance of random variable X nth order moment of random variable X nth order central moment of random variable X probability density function of the random variable X cumulative distribution function of real random variable X complementary cumulative distribution function (tail distribution) of real random variable X nth standardized cumulant of real random variable X ball of radius r centered at x (in d dimension) Rd /B(x, r ) indicator function covariance of two random variable X, Y the origin (of Rd ) usually designated to the cell administered by the BS located at o C B(o, r ) C B(o, r ) Wasserstein distance between the distribution of two random variables X and Y dx dx cumulative distribution function of a standard Gaussian random variable complementary cumulative distribution function of a standard Gaussian random variable (x) limxa f g (x) = 1,
Table 1.1: Mathematical notations.
f (x) g(x) as x a
f (x) = (g(x)) as x a
0 < lim inf xa
f (x) g (x)
lim supxa
f (x) g (x)
<
Abbreviation LHS RHS PDF CDF CCDF BS
Explanation Left hand side Right hand side Probability density function Cumulative distribution function Cumulative distribution function Base station
Table 1.2: Abbreviations
16
1. Introduction
17
Part I
Mathematical background and contributions
19
Chapter 2
Poisson point process

Contents
2.1 2.2 2.3 2.4 2.5 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 The use of Poisson point process on wireless modelling . . . . 19 Poisson point process . . . . . . . . . . . . . . . . . . . . . . . . . 21 Malliavin calculus on Poisson point process . . . . . . . . . . . . 23 Distribution of linear functional of Poisson point process . . . 26 2.5.1 2.5.2 2.5.3 2.6 Moments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Existing results on Gaussian approximation, Edgeworth expansion and concentration inequality . . . . . . . . . . . . . . . . . . New results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 28 31
Conclusion
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
2.1
Introduction
In this chapter, we review main mathematical results used in this thesis. First we give the denition Poisson measure on Polish space. From this denition, we derive useful results concerning Poisson point process. We study the distribution of linear functional of Poisson point process. We introduce some basics of Malliavin calculus on Poisson space then present results on a large class of functional of Poisson point process. Materials in this chapter are collected from literature, mainly from [8], [9], [10], [11], [12], [13], [14], [15], [16], and we also add new results from our recent work [6].
2.2
The use of Poisson point process on wireless modelling
In this section, we explain why and how stochastic geometry in general and spatial Poisson point processes in particular can help modeling the wireless communication systems. We also give a brief detail on Malliavin calculus for Poisson point process. More than a century ago, the rst wireless telegraph system was invented. Nowadays, wireless communication systems consist of cellular networks (as described in the previous section), ad hoc networks, sensor networks, wi- networks and others that are applied to all aspects of life. In contrary to wired systems, in wireless systems there is a much greater amount of uncertainty and randomness on node locations or channel conditions and usually
20
2. Poisson point process Realistic modeling
Mathematical tractability
Figure 2.1: Trade-o between tractability and realistic modeling.
there is interference because many nodes share the same medium. A conguration of active users is never xed in time: nodes can move or simply change their state. The received signal depends on the distance between the transmitter and the receiver, and the channels condition called fading, can vary from time to time. Models for wireless communication systems must capture all these uncertainty and randomness. However, it is dicult to model perfectly a system: the more realistic a model is, the less tractable it is. There is always a balance between realistic modeling and tractability (illustrated in gure 2.1). Poisson point process is widely used because it has the largest arsenal of results. It is more simple to deal with than other point process model and more importantly, it is more tractable. This modeling consists of treating the given architecture of the network as random and analyzing it in a statistical way. Determining the network performance with deterministic conguration of nodes is much more dicult. Poisson point process, on the other hand, allows one to characterize the average behavior over many spatial realizations (snapshot) of a network. It also used to model time varying conguration of nodes and mobility, for the rst time, in this thesis. Since the works [17], [18], Poisson point processes are the basis of stochastic geometry modeling of large wireless communication systems. It is successfully used to characterize the distribution of interference, outage probability, transport capacity, connectivity or delay in large ad hoc networks [19],[20], [21], [22], [23], etc, see also [24] or [25] for an overview. It is also used to study the performance and optimization of sensor networks [26], [27], etc . Poisson point process is applied on cellular networks as in [28], [17], [29], etc. Poisson point process, together with Malliavin calculus and Stein method, has allowed us to obtain theoretical results [14], [16], [9], [13]. However the application of this modern mathematics on wireless communication networks is very limited. Only recently, it was applied to solve the problem of full coverage on sensor networks, see [30]. The basic tools of Malliavin calculus consist in a gradient and a divergence operator which are linked by an integration by parts formula. For a Poisson point process w in a space E and a functional F depending on w, the gradient operator Dy can be dened by Dy F (w) = F (w y ) F (w) with y E . A basic of the stochastic analysis of Poisson point process is that for a large class of random variable F , the chaos representation of F is guaranteed: F = 1 In (fn ) n! n=0
where In is the multiple Poisson integral and the function fn is computed thanks to the gradient operator. fn (y1 , ..., yn ) = E [Dy1 ...Dyn F ]
21 Interestingly, this is similar to the way as Taylor expansion in analysis is established: g(x) =
n=0
an
xn where n!
an =
ng (0). xn
Here we have the correspondence: Analysis f (x) f (0)

ng n x ng xn (0)
Stochastic analysis F E [F ] Dn E [D n F ]
Basics and results on Poisson point process are presented in details in the next chapter.
2.3
Poisson point process
Let E be a compact metric space (i.e E can be partitioned into a countable union of compact metric spaces) with a diuse Radon measure . The space of congurations of E is the set of locally nite simple point measures E := w = zi (at most countable) , zi E ,
where z denotes the Dirac measure at z E , i.e z (A) = 1{z A} , A B (E ) Here, simple measure means that w({z }) 1 and locally nite measure means that w(K ) < for all compact K E . The conguration space E is endowed with the vague topology and its associated -algebra denoted by F E . For convenience, it is quite often to identify an element w = zi with its corresponding support, i.e the unordered set {z1 , ..., zn }, n N {+}. Also, w(A) counts the number of points in A B (E ). The distribution of point process w is characterized by the family of nite dimensional distributions (w(A1 ), ..., w(An )) where A1 , ..., An are mutually disjoint compact subsets of E . Denition 1. w is a Poisson point process (PPP) of intensity if for all set (A1 , ..., An ) of mutually disjoint compact subsets of E :
n
P(w(A1 ) = k1 , ..., w(An ) = kn ) =

i=1
e (Ai )
( (Ai ))ki ki !
If E = Rd ; B is Borel algebra of Rd and ( d z ) = d z , we will call w the homogenous Poisson point process with intensity parameter on Rd . Roughly speaking, the number of points of w falling into a subset A follows Poisson distribution of parameter (A), and the number of points falling into 2 disjoint subsets are independent. In what follows, we describe some operations on point process that preserves the Poisson law ([8]).
22
2. Poisson point process
0.9
0.9
0.8
0.8
0.7
0.7
0.6
0.6
0.5
0.5
0.4
0.4
0.3
0.3
0.2
0.2
0.1
0.1
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Figure 2.2: A realization of homogenous Poisson point process of intensity = 2.103 and its thinning with p(z ) = 0.2.
(Superposition) For n Poisson point process w1 , ..., wn of intensities 1 , ..., n (n < ), we call w = n i=1 wi their superposition. It is well known that w is a Poisson point process with intensity = n i=1 i . (Thinning) Consider a Poisson point process w of intensity and a function p : E [0, 1]. The thinning of w with retention function p is given by wp = i zi where the random variables {i }i are independent given w and P(i = 1|w) = p(zi ) = 1 P(i = 0|w). If p is measurable then wp is a Poisson point process of intensity p with p (A) = A p(z ) d (z ). (Transformation, displacement theorem) Consider another compact metric space E and a probability kernel p(z, .), i.e for all z E , p(z, .) is a probability measure in E . The transformation of a Poisson point process w by p with intensity in where z , z , ... are independent given w and has the E is dened as wp = zi 1 2 probability P(zi A |w) = p(zi , A ). It is shown that wp is a Poisson point process of intensity (A ) = E p(z, A ) d (z ). Consider a second compact metric space E . For each point zi of w, we associate with a random mark mi E . The mark point process w can be presented as w=
i
(zi ,mi ) ,
where is the Dirac measure on E E . w is said to be independent marked if the marks are mutually independent given a realization of w and the distribution of marks depends only on the location of its parent point P(m .|w) = P(m .|z ) = d Kz (m) with a probability kernel K. (.) from E to E . By displacement theorem we can prove that if w is a Poisson point processwith intensity , then w is a Poisson point processin E E with intensity measure: (A A ) =
A A
d Kz (m) d (z ),
A E, A E
If additionally the marks are independent of their parent point, i.e d Mz (m) is the same as d M (m) then w is a Poisson point processin E E with intensity d (z ) d K (m).
23
2.4
Malliavin calculus on Poisson point process
In this section we introduce basics of Malliavin calculus on Poisson space and then present some results on general functional depending on a Poisson point process. A real function f : E n R is called symmetric if f (z(1) , , z(n) ) = f (z1 , , zn ) for all permutations of n . The space of symmetric square integrable random variables with respect to is denoted by L2 ( )n . Let n = {(z1 , , zn ) E n | zi = zj , i = j }. For f L2 ( )n , the multiple Poisson stochastic integral In (fn ) is then dened as In (fn )(w) =
n
fn (z1 , , zn )( d w(z1 ) d (z1 )) ( d w(zn ) d (zn ))
It is well known that, E

zi =zj w 1i<j n
If n = 1, (2.1) is the same as Campbells formula. From this, by introduction we can show that E [In (fn )] = 0. If fn L2 ( )n and gm L2 ( )m , n m, the following isometry formula holds: E [In (fn )Im (gm )] = n!1{m=n} fn , gm (see [9]). Here fn , gn
L2 ( )n L2 ( )n
f (z1 , , zn ) =
En
f (z1 , , zn ) d (z1 ) d (zn )
(2.1)
(2.2)
=
En
fn (z1 , ..., zn )gn (z1 , ..., zn ) d (z1 )... d (zn )
For fn L2 ( )n and gm L2 ( )m , we dene fn l k gm , 0 l k , to be the function: (yl+1 , , yn , xk+1 , , xm )

Yl
fn (y1 , , yn )gm (y1 , , yk , xk+1 , , xm ) d (y1 ) . . . d (yl ). (2.3)
l We denote by fn l k gm the symmetrization in n + m k l variables of fn k gm , 0 l k . This leads us to the next proposition, shown in [9]:
Proposition 1. For fn L2 ( )n and gm L2 ( )m , we have In (fn )Im (gm ) = where hn,m,s =

s2i2(snm) 2(nm) s=0
In+ms (hn,m,s ),
i!
n i
m i
i s i fn i gm si
belongs to L2 ( )n+ms , 0 s 2(m n).
24
In what follows, given f L2 ( )q (q 2) and z E , we denote by f (, z ) the function on E q1 given by (x1 , , xq1 ) f (x1 , , xq1 , z ). Furthermore, we have: Theorem 2. Every random variable F L2 (E , P) admits a (unique) Wiener-Poisson decomposition of the type F = E [F ] +
n=1
In (fn ),
where the series converges in L2 (E , P) and, for each n 1, fn is an element of L2 ( )n . Moreover, we have the isometry equation: V [F ] =
n=1
n! fn
2 L2 ( )n
(2.4)
Denition 2. Let domD be the the set of random variables F L2 (E , P) admitting a chaotic decomposition such that
n=1
n.n! fn
2 L2 ( )n
< .
Let D be dened by D : domD L2 (E E, P ), such that F = E [F ] +

n 1
In (fn ) Dz F () =
n 1
nIn1 (fn (, z )).
Theorem 3. We have Dz F (w) = F (w + z ) F (w), d P d a.e. The theorem 3 says that, the operator D is nothing else but simply the impact made on F by adding a point z on w. The integration by parts then says that, for any F domD , any u L2 ( ), E F
E
u(z )( d w(z ) d (z )) = E
Dz F u(z ) d (z ) .
E
(2.5)
It is well known that F is a linear functional (see the next section) F = E f d w if and only if the chaos representation of F is F = E [F ] + I1 (f ) and in this case Dz F = f (z ) and thus V [F ] = E |Dz F |2 d z . Generally, F has the form
k
F =
i=1 (z1 ,...,zi)i ;zj w
fi (z1 , ..., zi )
if and only if the chaos representation of F has the form

k
F = E [F ] +
i=1
Ii (gi )
where g1 , ..., gk can be derived from f1 , ..., fk . Using the dierence operator D , we can bound the variance of F as follows:
25 Theorem 4. ([9], [14]) F L2 (E , P) we have: V [F ] E Equality occurs if and only if F is linear. Corollary 1. F L2 (E , P) such that |Dz F | f (z ) for some non negative measurable function f : E R for all z, w then: V [F ]
E E
|Dz F |2 d z
(2.6)
f 2 (z ) d (z )
(2.7)
Theorem 5. ([9],[31]) Consider two functionals F, G L2 (E , P). Assume that Dz F, Dz G 0, P a.s. then C [F, G] 0. As a consequence, V [F + G] V [F ] + V [G]. Denition 3. The Ornstein-Uhlenbeck operator L is given by LF =
n=1
nIn (fn ),
whenever F domL, given by those F L2 ( )n such that their chaotic expansion veries
n=1
n2 .n! fn
2 L2 ( )n
< .
Note that E [LF ] = 0, by denition and (2.2). Denition 4. For F L2 (P) such that E [F ] = 0, we may dene L1 by L
1
F =
n=1
1 In (fn ). n
The three Malliavin-type operators dened above give us way derive upper bounds on the distribution of a functional F (theorem 7) and an estimate of error on Gaussian approximation (theorem 6), results which is more general than the ones introduced on the previous sections. Theorem 6. ([13]) Let F domD be such that E [F ] = 0 and V [F ] = 1. Then, dW (F, N (0, 1)) E 1 [Dz F Dz L1 F ] d (z ) +
E
E |Dz F |2 Dz L1 F
d (z )
26
Theorem 7. [14], Let F domD . Assume that |DF | M , P a.s., for some M 0 and E |Dz F |2 d (z ) 2 , Pa.s. Then for all u > 0 we have P(F E [F ] u) exp for M > 0 and P(F E [F ] a) exp for M = 0 where g(t) = (1 + t) ln(1 + t) t. Corollary 2. Let F domD . Assume that |DF | M , P a.s., for some M > 0 and |Dz F f (z )|, Pa.s. for some non negative function f L2 ( ). Then for all u > 0 we have P(F E [F ] u) exp
E
2 g M2
u.M 2
(2.8)
2 2u2
(2.9)
f 2 (z ) d (z ) g M2
f 2 (z )
uM d (z )
(2.10)
2.5
Distribution of linear functional of Poisson point process
We call F a linear functional of w if there exists f : E R such that F =

E
f (z ) d w(z ) =
z w
f (z )
We assume that f L1 ( ). In this section we are interested in the distribution of F . Let Lw (.) be the Laplace functional of w, i.e: Lw (u) = E e
E
u(z ) d w (z )
(2.11)
Theorem 8. ([9]) The Laplace functional of w satises: Lw (u) = e

E (1e u (z ) )
d (z )
u L1 ( )
(2.12)
From the above theorem the moment generating function (MGF) of F is expressed as follows: f (z ) E e E f d w = e E (e 1) d (z ) (2.13)
2.5.1
Moments
The complete Bell polynomials Bn (a1 , ..., an ) are dened as follows: exp an n n! n=1
Bn (a1 , a2 , ..., an ) n n! n=1
for all a1 , ..., an and such that all above terms are correctly dened.
27 The rst ve Bell complete polynomials are given as: B1 (a1 ) = a1 B2 (a1 , a2 ) = a2 1 + a2 B3 (a1 , a2 , a3 ) = a3 1 + 3a1 a2 + a3
2 2 B4 (a1 , a2 , a3 , a4 ) = a4 1 + 4a1 a2 + 4a1 a3 + 3a2 + a4 3 2 2 B5 (a1 , a2 , a3 , a4 , a5 ) = a5 1 + 10a1 a2 + 10a1 a3 + 15a1 a2 + 5a1 a4 + 10a2 a3 + a5
It is well known that the coecients of Bells polynomial are always non negative. If a S random variable S has n rst cumulants S 1 , ..., n then its moments and central moments of S can be expressed as:
S mn [S ] = E [S n ] = Bn (S 1 , ..., n ) S cn [S ] = E [(S E[S ])n ] = Bn (0, S 2 ..., n )
Now consider a linear functional F of w, i.e F = E f d w for some non-negative measurable function f . We can compute the cumulants, the moments and central moments of F by Bell complete polynomials as follows:
i Theorem 9 (Generalization of Campbells formulas). Assume that f n i=1 L (E, ). The F i cumulants of F = E f d w is i = E f (z ) d (z ) (i = 1..n). The moments and central moments of F are given as:
mi [F ] = Bi
E
f (z ) d (z ),
E
f 2 (z ) d (z ), ...,
E
f i (z ) d (z )
(2.14)
and ci [F ] = Bi 0,
E
f 2 (z ) d (z ), ...,
E
f i (z ) d (z )
(2.15)
for i = 1, 2, ..., n. Proof. We apply the theorem 8 to get that: E eF =e =e =e

E
(ef (z ) 1) d (z )
n=1
n f n (z ) d (z )
E
n=1
f n (z ) d (z )
i By denition of cumulant, we have F i = E f (z ) d (z ), and thus the expressions for moments and central moments of F are straightforward.
As a direct consequence, one can easily obtain from the above theorem two useful formulas (Campbell): Corollary 3. Let F =
E
f d w then E [F ] =
E
f (z ) d (z ), f L1 ( ) f 2 (z ) d (z ), f L2 ( )
V [F ] = E (F E [F ])
=
E
28
Notice that if we consider a independent making Poisson point process w = (zi ,mi ) with a probability kernel K from E to the space of marks E and F = f (zi , mi ) then f i (z, m) d Kz (m) d (z ) =
E
E f i (z, m) d (z )
Campbells formula is written: E [F ] =

E
E [f (z, m)] d (z ), E f 2 (z, m) d (z )
V [F ] =
E
2.5.2
Existing results on Gaussian approximation, Edgeworth expansion and concentration inequality
We have now expressions for moments and central moments of F . We note that the central moments of F is always non negative as f is supposed to be non negative. One can ask if there is mean to compute the tail distribution of F . Gaussian approximation seems to be a rst answer one can think of due to the central limit theorem (CLT). An error bound of Gaussian approximation for sum of n i.i.d random variables is known as BerryEsseen theorem. It is possible to nd an alternative version of this error bound 2 a for linear functional of PPP. Let Q(a) = 1 eu /2 d u be the CDF of a standard 2 Gaussian random variable and Q be its CCDF. Theorem 10. Consider F =
E
f d w with f L2 ( ). Let F =
E
F E[F ] V [F ] ,
then: (2.16)
P(F a) Q(a)
|f (z )|3 d (z ) ( E f 2 (z ) d (z ))3/2
Proof. Unfortunately there is no elementary proof for this theorem, we must apply the theorem 6 for F . First, note that the chaos presentation of F is F = I1 ( Vf [F ] ). Then Dz F = Dz L1 F = Thus, V [F ] = get (2.16). f (z ) V [F ]
f V [F ] .
f (z ) d (z ) and Dz L1 F = Dz F =
We now apply theorem 6 to
Now instead of d (z ), we consider d (z ) in the above theorem, and let F , F the corresponding versions of F, F , we get: P(F a) Q(a)
E
|f (z )|3 d (z )
f 2 (z ) d (z ))3/2
(2.17)
1 ). The last inequality shows that the error of Gaussian approximation is around O( It is worth noted that the error bound for Gaussian approximation of the sum of n i.i.d 1 ). variables is O( n
29 An improvement for Gaussian approximation is known as Edgeworth expansion. Let F n = F n = (V [F ])n/2 ( f n (z ) d (z ) 2 n/2 E f (z ) d (z ))
E
be the nth standardized cumulant of F . The Edgeworth expansion for the distribution of F is given as: P(F a) = Q(a) +
n=1
Pn (D )Q(a)
(2.18)
where Pn is polynomial of degree 3n and D is the dierential operator. The rst ve terms are: P(F > a) = Q(a) + F + 3 Q(3) (a) 6 F (F )2 4 Q(4) (a) + 3 Q(6) (a) + 24 72 F F F (F )3 5 Q(5)(a) + 3 4 Q(7) (a) + 3 Q(9) (a) + 120 144 1296 2 F F F F ( ) (F )2 F (F )4 6 (3) 4 Q (a) + ( 4 + 3 5 )Q(8) (a) + 3 Q(10) (a) + 3 Q(12) (a) + 720 1152 720 1782 31104 ... For the best of our knowledge, no error bound for Edgeworth expansions exists in the literature. Even we cannot nd any reference in Edgeworth expansions of linear functional of Poisson point process. However, if f (z ) 0 we can rewrite (2.17) as: P(F a) Q(a) F 3 (2.19) This provides a hint that, error bounds may be expressed as function of cumulants of F under appreciate conditions of f . We have presented the Gaussian approximation and Edgeworth expansions for linear functionals. We are now interested in upper bounds on the distribution of F , which can be called concentration inequality. Theorem 11. Let M, a > 0. Assume that 0 f (z ) M a.s and f L2 (E, ), then: P(F > E [F ] + a) exp E [F ] g M a E [F ] (2.20)
where g(u) = (1 + u) ln(1 + u) u. Assume that |f (z )| M a.s and f L2 (E, ), then P(F > E [F ] + a) exp V [F ] g M2 a.M V [F ] (2.21)
Assume that f (z ) 0 a.s and f L2 (E, ), then P(F > E [F ] + a) exp V [F ] 2a2 (2.22)
30
The above theorem can be directly derived from 2, which will be introduced in the next section. However let us take this opportunity to prove this theorem in a very nice, simple and elementary fashion, exactly the same way as Bennett built his concentration inequality for the sum of n i.i.d random variables. Proof. Using Chernos bound we have: P(F > E [F ] + a) E eF /e(E[F ]+a)
f (z ) 1f (z ) d (z )a ) = e E (e
Now assume that 0 f (z ) M a.s . Observe that the function f (z ) 1 eM 1 R+ , we have e f (z for f (z ) = 0, thus ) M ef (z ) 1 a.s . We deduce that: P(F > E [F ] + a) exp = exp
E
e x 1 x
is increasing on
eM 1 f (z ) M
eM
eM 1 f (z ) f (z ) d (z ) a M 1 M E [F ] a M
We minimize the L.H.S in , by some elementary manipulations, we reach the optimal 1 a value = M ln 1 + E[ F ] and we obtain (2.20). Now assume that |f (z )| M a.s . Observe that the function on R (the value at 0 is 1/2), we have that ef (z ) f (z ) 1 a.s . Thus, P(F > E [F ] + a) exp = exp
E eM ex 1x x2
is increasing
eM 1 M 2 f (z ) M2
eM M 1 2 f (z ) d (z ) a M2 1 M V [F ] a M2
aM 1 ln 1 + V minimizes the L.H.S and apply this value we obtain We nd that = M [F ] (2.21). Now assume that f (z ) 0 a.s . By the same argument as above
ef (z ) f (z ) 1 Thus, P(F > E [F ] + a) exp = exp Minimizing the LHS by =

a V [F ]
2 2 f (z ) 2
2 f 2 (z ) d (z ) a 2 E 2 V [F ] a 2
we obtain (2.22).
31
2.5.3
New results
In this subsection we present our new results on the distribution of linear functional F . We consider the measure d (z ) instead of d (z ) and we are interested in F and F and let F m(p, ) = n . We denote by = f L2 ( ) and f = f / . Note that f L2 ( ) = 1/ and that p 1p/2 . |f (z )|p d (z ) = f f p m(p, ) := Lp ( ) L2 ( )
E
For > 0, let N =

E
f (x)( d w(z ) d (z )) = F E F .
In what follows, we consider G(N ) where G : R R. We note that if G = 1{tT } then P(N T ) = E G(N ) . The proof of the following theorem may be found in [32],[9] or [13]: Theorem 12. Let f L2 ( ). Then, for any Lipschitz function G from R to R, we have E G(N )
3 (R, R), Theorem 13. For G Cb R
G d
1 2
m(3, ) G 2
Lip .
E G(N )
G(y ) d (y )
1 m(3, ) 6
G(y )H3 (y ) d (y )
R
m(4, 1) m(3, 1)2 + 6 9
G(3)
(2.23)
Proof. According to the Taylor formula, Dx G(N ) = G(N + f (x)) G(N )
1 2 1 = G (N )f (x) + f (x) G (N ) + f (x)3 2 2 Hence, according to (2.5) and (2.24), E N (Pt G) (N ) = E

E
1 0
r 2 G(3) (rN + (1 r )f (x)) d r. (2.24)
f (x)Dx (Pt G) (N ) d (x) 1 2

3 f (x) d (x)E (Pt G)(3) (N ) 1 0
= E (Pt G) (N ) + +
1 f 4 (x) d (x)E 2 E = A1 + A2 + A3 .
(Pt G)(4) (rN + (1 r )f (x))r 2 d r
Hence, e4t |A3 | 6 1 e2t 2 m(4, ) G(3)

.
32 Moreover, according to Theorem 12, E (Pt G)(3) (N ) (Pt G)(3) (x) d (x)
R
1 m(3, ) (Pt G)(4) 2 2 1 e4t m(3, ) G(3) . 2 1 e2t
Then, we have, 1 |A2 m(3, ) 2 Hence, 1 E N (Pt G) (N ) (Pt G) (N ) = m(3, ) 2 where R(t) Now then, (Pt G)(3) (x) d (x) = e3t
R R R
(Pt G)(3) (x) d (x)|
1 e4t G(3) m(3, )2 2 t 4 1e
(Pt G)(3) (x) d (x) + R(t),

R
m(3, )2 m(4, ) + 4 6
G(3)
e4t 1 e2t
G(3) (et x + G(3) (y ) d (y )

R
1 e2t y ) d (y )
= e3t = e3t
R
G(y )H3 (y ) d (y ),
since the Gaussian measure on R2 is rotation invariant and according to (10.2). Remarking that
0
e4t (1 e2t )1/2 d t = 2/3
and applying (10.4) to x =
N ,
the result follows.
This development is not new in itself but to the best of our knowledge, it is the rst time that there is an estimate of the error bound. Following the same lines, we can pursue the expansion up to any order provided that F be suciently dierentiable. Namely, for 5 , we have F Cb E G(N ) =
R
G(y ) d (y ) + + m(4, 1) 24
m(3, 1) 6
G(3) (y ) d (y ) +
R .
m(3, 1)2 72
G(6) (y ) d (y )
R
G(4) (y ) d (y ) + G G(5)
R
(2.25)
where G m(3, 1) 3/2 4 2 2 m(3, 1)2 + ( + ) 45 135 128 2 m(4, 1) . (2.26)
33
2.6
Conclusion
In this chapter we introduce all the mathematical tools used throughout this thesis. First we have presented the notion of Poisson point process. After, we have introduced the notation of Malliavin calculus applied to Poisson point process and presented useful results, which are upper bounds on the distribution of random variables depending on a Poisson point process. Then we have studied the distribution of linear functional of Poisson point process, we have presented some new results on Edgeworth expansion.
34
35
Part II
Dimensioning and coverage models
37
Chapter 3
Robust methods for LTE and WiMAX dimensioning

Contents
3.1 3.2 3.3 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 System Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39 Loss probability . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 3.3.1 3.3.2 3.3.3 3.4 Exact method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Approximations . . . . . . . . . . . . . . . . . . . . . . . . . . . . Robust upper-bound . . . . . . . . . . . . . . . . . . . . . . . . . 42 42 44
Applications to OFDMA and LTE . . . . . . . . . . . . . . . . . 44
3.1
Introduction
Future wireless systems will widely rely on OFDMA (Orthogonal Frequency Division Multiple Access) multiple access technique. OFDMA can satisfy end users demands in terms of throughput. It also fullls operators requirements in terms of capacity for high data rate services. Systems such as 802.16e and 3G-LTE (Third Generation Long Term Evolution) already use OFDMA on the downlink. Dimensioning of OFDMA systems is then of the utmost importance for wireless telecommunications industry. OFDM (Orthogonal Frequency Division Multiplex) is a multi carrier technique especially designed for high data rate services. It divides the spectrum in a large number of frequency bands called (orthogonal) subcarriers that overlap partially in order to reduce spectrum occupation. Each subcarrier has a small bandwidth compared to the coherence bandwidth of the channel in order to mitigate frequency selective fading. User data is then transmitted in parallel on each sub carrier. In OFDM systems, all available subcarriers are aected to one user at a given time for transmission. OFDMA extends OFDM by making it possible to share dynamically the available subcarriers between dierent users. In that sense, it can then be seen as multiple access technique that both combines FDMA and TDMA features. OFDMA can also be possibly combined with multiple antenna (MIMO) technology to improve either quality or capacity of systems. In practical systems, such as WiMAX or 3G-LTE, subcarriers are not allocated individually for implementation reasons mainly inherent to the scheduler design and physical
38
3. Robust methods for LTE and WiMAX dimensioning
Figure 3.1: OFDMA principle : subcarriers are allocated according to the required transmission rate
layer signaling. Several subcarriers are then grouped in subchannels according to dierent strategies specic to each system. In OFDMA systems, the unit of resource allocation is mainly the subchannels. The number of subchannels required by a user depends on his channels quality and the required bit rate. If the number of demanded subchannels by all users in the cell is greater than the available number of subchannel, the system is overloaded and suer packet losses. The questions addressed here can then be stated as follows: how many subchannels must be assigned to a BS to ensure a small overloading probability ? Given the number of available subchannels, what is the maximum load, in terms of mean number of customers per unit of surface, that can be tolerated ? Both questions rely on accurate estimations of the loss probability. The objectives of this chapter are twofold: First, construct and analyze a general performance model for an isolated cell equipped with an OFDMA system as described above. We allows several classes of customers distinguished by their transmission rate and we take into account path-loss with shadowing. We then show that for a Poissonian conguration of users in the cell, the required number subchannels follows a compound Poisson distribution. The second objective is to compare dierent numerical methods to solve the dimensioning problem. In fact, there exists an algorithmic approach which gives the exact result potentially with huge memory consumption. On the other hand, we use and even extend some recent results on functional inequalities for Poisson processes to derive some approximations formulas which turn to be rather eective at a very low cost. When it comes to evaluate the performance of a network, the quality of such a work may be judged according to several criteria. First and foremost, the exactness is the most used criterion: it means that given the exact values of the parameters, the real system, the performances of which may be estimated by simulation, behaves as close as possible to the computed behavior. The sources of errors are of three kinds: The mathematical model may be too rough to take into account important phenomena which alter the performances of the system, this is known as the epistemic risk. Another source may be in the mathematical
39 resolution of the model where we may be forced to use approximate algorithms to nd some numerical values. The third source lies in the lack of precision in the determination of the parameters characterizing the system: They may be hard, if not impossible, to measure with the desired accuracy. It is thus our point of view that exactness of performance analysis is not all the matter of the problem, we must also be able to provide condence intervals and robust analysis. That is why, we insist on error bounds in our approximations. Resources allocation on OFDMA systems have been extensively studied over the last decade, often with joint power and subcarriers allocation, see for instance [33, 34, 35, 36]. The problem of OFDMA planning and dimensioning have been more recently under investigation. In [37], the authors propose a dimensioning of OFDMA systems focusing on link outage but not on the other parameters of the systems. In [38], the authors give a general methodology for the dimensioning of OFDMA systems, which mixes a simulation based determination of the distribution of the signal-to-interference-plus-noise ratio (SINR) and a Markov chain analysis of the trac. In [39, 40], the authors propose a dimensioning method for OFDMA systems using Erlangs loss model and Kaufman-Roberts recursion algorithm. In [41], the authors study the eect of Rayleigh fading on the performance of OFDMA networks. The article is organized as follows. In Section 3.2, we describe the system model and set up the problem. In Section 3.3, we examine four methods to derive an exact, approximate or robust value of the number of subchannels necessary to ensure a given loss probability. In Section 3.4, we apply these formulas to the particular situation of OFDMA systems.
3.2
System Model
In practical systems, such as WiMAX or 3G-LTE, resource allocation algorithms work at subchannel level. The subcarriers are grouped into subchannels that the system allocates to dierent users according to their throughput demand and mobility pattern. For example, in WiMAX, there are three modes available for building subchannels: FUSC (Fully Partial Usage of Subchannels), PUSC (Partial Usage of SubChannels) and AMC (Adaptive modulation and coding). In FUSC, subchannels are made of subcarriers spread over all the frequency band. This mode is generally more adapted to mobile users. In AMC, the subcarriers of a subchannel are adjacent instead of being uniformly distributed over the spectrum. AMC is more adapted to nomadic or stationary users and generally provides higher capacity. The grouping of subcarriers into subchannels raises the problem of the estimation of the quality of a subchannel. Theoretically channel quality should be evaluated on each subcarrier of the corresponding subchannel to compute the associated capacity. This work assumes that it is possible to consider a single channel gain for all the subcarriers making part of a subchannel (for example via channel gains evaluated on pilot subcarriers). We consider a circular cell C of radius R with a base station (BS for short) at its center. The transmission power dedicated to each subchannel by the base station is denoted by P . Each subchannel has a total bandwidth W (in kHz). The received signal power for a mobile station at distance d from the BS can be expressed as P (d) = P K GF := P Gd , d (3.1)
where K is a constant equal to the attenuation at a reference distance, denoted by dref ,
40
that separates far eld from near eld propagation. Namely, K = c 4f dref
2
d ref ,
where f is the radio-wave frequency. The variable is the path-loss exponent which indicates the power at which the path loss increases with distance. Its depends on the specic propagation environment, in urban area, it is in the range from 3 to 5. It must be noted that this propagation model is an approximate model, dicult to calibrate for real life situations. In particular, it might be reasonable to envision models where depends on the distance so that the attenuation would be proportional to d (d) . Because of the complexity of such a model, is often considered as constant but the path-loss is multiplied by two random variables G and F which represent respectively the shadowing, i.e. the attenuation due to obstacles, and the Rayleigh fading, i.e. the attenuation due to local movements of the mobile. Usually, G is taken as a log-normal distribution: G = 10S/10 , where S N (, v 2 ). As to F , it is customary to choose an exponential distribution with parameter 1. Both, the shadowing and the fading experienced by each user are supposed to be independent from other users shadowing and fading. For the sake of simplicity, we will here treat the situation where only shadowing is taken into account, the computations would be pretty much like the forthcoming ones and the results rather similar should we consider Rayleigh fading. All active users in the cell compete to have access to some of the Navail available subchannels. There are K classes of users distinguished by the transmission rate they require: Ck is the rate of class k customers and k denotes the probability that a customer belongs to class k. A user, at distance d from the BS, is able to receive the signal only if d) is above some constant min where the signal-to-interference-plus-noise ratio SNR = P ( I I is the noise plus interference power and P (d) is the received signal power at distance d, see (3.1). If the SNR is below the critical threshold, then the user is said to be in outage and cannot proceed with his communication. To avoid excess demands, the operator may impose a maximum number Nmax of allocated subchannels to each user at each time slot. According to the Shannon formula, for a user demanding a service of bit rate Ck , located at distance d from the BS and experiencing a shadowing g, the number of requires subchannels is thus the minimum of Nmax and of Ck if P gd /I min , W log2 (1 + P gd /I ) Nuser = 0 otherwise,
where x means the minimum integer number not smaller than x. We make the simplifying assumption that the allocation is made at every time slot and that there is no buering neither in the access point nor in each mobile station. All the users have independently from others a probability p to have a packet to transmit at each slot. This means, that each user has a trac pattern which follows a geometric process of intensity p. We also assume that users are dispatched in the cell according to a Poisson process of intensity 0 . According to the thinning theorem for Poisson processes, this induces that active users form a Poisson process of intensity = 0 p. This intensity is kept xed over the time. That may result from two hypothesis: Either we consider that for a small time scale, users do not move signicantly and thus the conguration does not evolve. Alternatively, we may consider that statistically, the whole conguration of active
41 users has reached its equilibrium so that the distribution of active users does not vary through time though each user may move. From the previous considerations, a user is characterized by three independent parameters: his position, his class and the intensity of the shadowing he is experiencing. We model this as a Poisson process on E = B (0, R) {1, , K } R+ of intensity measure where B (0, R) = {x R2 , x R}, is the probability distribution of classes given by ({k}) = k and is the distribution of the random variable G dened above. We set f (x, k, g) = min Nmax , 1{P g With the notations of Section ??, Ntot =
cell x
I min }
d (x) := ( dx d (k) d(g))
Ck W log2 (1 + P g x
/I )
f (x, k, g) d (x, k, g).
We are interested in the loss probability which is given by P(Ntot Navail ). We rst need to compute the dierent moment of f with respect to in order to apply Theorem 12 and Theorem 13. For, we set lk = Nmax I P Ck , W log2 (1 + min )
where a b = min(a, b). Furthermore, we introduce k, 0 = , k, l = 2Ck /W l 1 , 1 k K, 1 l lk 1,
and k, lk = Imin /P. By the very denition of the ceiling function, we have
K lk
f d =
E k =1
k
l=1
lp
cell R
1[k, l ; k, l1 ) (g x
) d(g) dx.
According to the change of variable formula, we have

cell
1[k, l ; k, l1 ) (g x
) dx = (k, l
2/
R2 k, l1 R2 )g2/ .
2/
Thus, we have 1[k, l ; k, l1 ) (g x
) d(g) dx = (k, l = (k, l

2/
cell
R2 k, l1 R2 )E 10S/5 R2 k, l1 R2 ) 10
2/
v ln 10)/5 (+ 10 2
2/
2/
:= k, l .
We thus have proved the following theorem. Theorem 14. For any p 0, with the same notations as above, we have:
K lk
f p d =
k =1
k
l=1
lp k, l .
(3.2)
42
3.3
3.3.1
Loss probability
Exact method
Since f is deterministic, Ntot follows a compound Poisson distribution: it is distributed as

K lk
l Nk, l
k =1 l=1
where (Nk, l , 1 k K, 1 l lk ) are independent Poisson random variables, the parameter of Nk, l is k k, l . Using the properties of Poisson random variables, we can reduce the complexity of this expression. Let L = max(lk , 1 k K ) and for l {1, , L}, let Kl = {k, lk l}. Then, Ntot is distributed as
L
l Ml
l=1
where (Ml , 1 l lk ) are independent Poisson random variables, the parameter of Ml being ml := kKl k k, l . For each l, it is easy to construct an array which represents the distribution of lMl by the following rule: pl (w) = 0 exp(ml )mq l /q ! if w mod l = 0, if w = ql.
By discrete convolution, the distribution of Ntot and then its cumulative distribution function, are easily calculable. The value of Navail which ensures a loss probability below the desired threshold is found by inspection. The only diculty with this approach is to determine where to truncate the Poisson distribution functions for machine representation. According to large deviation theory [42], P(Poisson( ) a ) exp( (a ln a + 1 a)). When is known, it is straightforward to choose a( ) so that the right-hand-side of the previous equation is smaller than the desired threshold. The total memory size is thus proportional to max(ml a(ml )l, 1 l lk ). This may be memory (and time) consuming if the parameters of some Poisson random variables or the threshold are small. This method is well suited to estimate loss probability since it gives exact results within a reasonable amount of time but it is less useful for dimensioning purpose. Given Navail , if we seek for the value of which guarantees a loss probability less than the desired threshold, there is no better way than trial and error. At least, the subsequent methods even imprecise may help to evaluate the order of magnitude of for the rst trial.
3.3.2
Approximations
We begin by the classical Gaussian approximation. It is clear that P(

E
f d Navail ) =
P(
f ( d d ) N )
E
= E 1[N , +) (
f ( d d ))
43 where N = (Navail f d )/ . Since the indicator function 1[N , +) is not Lipschitz, we can not apply the bound given by Theorem 12. However, we can upper-bound the indicator by a continuous function whose Lipschitz norm is not greater than 1. For instance, taking (x) = min(x+ , 1) and N (x) = (x N ), we have 1[N +1, +) N +1 1[N , +) N 1 1[N 1, +) . Hence, 1 Q(N + 1) 1 2 2 m(3, 1) P( f d Navail ) 1 Q(N 1) + 1 2 2 m(3, 1) , (3.3)
where Q is the cumulative distribution function of a standard Gaussian random variable. According to Theorem 13, one can proceed with a more accurate approximation. Via l such that polynomial interpolation, it is easy to construct a C 3 function N
l (3) (N ) l 1[N , +) 1 and 1[N +3.5, +) N
r such that and a function N r (3) ) (N r 1 and 1[N , +) N 1[N 3.5, +)
From (10.3), it follows that 1 Q(N + 3.5) m(3, 1) (3) Q (N + 3.5) E P( f d Navail ) 6 E m(3, 1) (3) Q (N 3.5) + E (3.4) 1 Q(N 3.5) + 6
where E is the right-hand-side of (8.10) with F (3) = 1. Going again one step further, following the same lines, according to (2.25), one can show that P(
E
f d Navail ) 1 Q(N 6.5) + m(3, 1) (3) m(3, 1)2 (5) Q (N 6.5) + Q (N 6.5) 72 6 m(4, 1) (3) + Q (N 6.5) + F (3.5) 24
where F is bounded above in (2.26). For all the approximations given above, for a xed value of Navail , an approximate value of can be obtained by solving numerically an equation in .
44
3.3.3
Robust upper-bound
If we seek for robustness and not precision, it may be interesting to consider the so-called concentration inequality. We remark that in the present context, f is non-negative and bounded by L = maxk lk so that we are in position to apply Theorem 2.21. We obtain that f 2 d aL f d + a) exp E 2 g( P( f d ) , (3.6) 2 L E E E f d where g(u) = (1 + u) ln(1 + u) u.
3.4
Applications to OFDMA and LTE
In such systems, there is a huge number of physical parameters with a wide range of variations, it is thus rather hard to explore the while variety of sensible scenarios. For illustration purposes, we chose a circular cell of radius R = 300 meters equipped with an isotropic antenna such that the transmitted power is 1 W and the reference distance is 10 meters. The mean number of active customers per unit of surface, denoted by , was chosen to vary between 0, 001 and 0.000 1, this corresponds to an average number of active customers varying from 3 to 30, a realistic value for the systems under consideration. The minimum SINR is 0.3 dB and the random variable S dened above is a centered Gaussian with variance equal to 10. There are two classes of customers, C1 = 1, 000 kb/s and C2 = 400 kb/s. It must be noted that our set of parameters is not universal but for the dierent scenarios we tested, the numerical facts we want to point out were always apparent. Since the time scale is of the order of a packet transmission time, the trac is dened as the mean number of required subchannels at each slot provided that the time unit is the slot duration, that is to say that the load is dened as = cell f d . Figure 3.2 shows, the loss probability may vary up to two orders of magnitude when the rate and the probability of each class change even if the mean rate k k Ck remains constant. Thus mean rate is not a sucient parameter to predict the performances of such a system. The load is neither a pertinent indicator as the computations show that the loads of the various scenarios diers from less than 3%. Comparatively, Figure 3.2 shows that variations of have tremendous eects on the loss probability: a change of a few percents of the value of induces a variation of several order of magnitude for the loss probability. It is not surprising that the loss probability increases as a function of : as increases, the radio propagation conditions worsen and for a given transmission rate, the number of necessary subchannels increases, generating overloading. Beyond a certain value of (apparently around 3.95 on Figure 3.2), the radio conditions are so harsh that a major part of the customers are in outage since they do not satisfy the SNR criterion any longer. We remark here that the critical value of is almost the same for all congurations of classes. Indeed, the critical value c of can be found by a simple reasoning: When < c , a class k customer uses less than the allowed 1/ lk subchannels because the radio conditions are good enough for k, j R for some j < lk so that the load increases with . For > c , all the k, l are lower than R and the larger , the wider the gap. Hence the number of customers in outage increases as increases and the load decreases. Thus, c inf {, s, ls 1 R} for s = arg maxk lk .
1/ 1/
45
Figure 3.2: Impact of and on the loss probability (Navail = 92, = 0.0001)
If we proceed this way for the data of Figure 3.2, we retrieve c = 3.95. This means that for a conservative dimensioning, in the absence of estimate of , computations may be done with this value of . For a threshold given by = 104 , we want to nd Navail such that P(Ntot Navail ) . As said earlier, the exact method gives the result at the price of a sometimes lengthy process. In view of 3.3, one could also search for such that 1 Q() + 1 2 2 m(3, ) = (3.7)
and then consider 1 + E f d + as an approximate value of Navail . Unfortunately and as was expected since the Gaussian approximation is likely to be valid for large values of , the corrective term in (3.7) is far too large (between 30 and 500 depending on ) for (3.7) to have a meaning. Hence, we must proceed as usual and nd such that 1 Q() = , i.e. 3.71. The approximate value of Navail is thus given by E f d + 3.71 . The consequence is that we do not have any longer any guarantee on the quality of this approximation, how close it is to the true value and even more basic, whether it is greater or lower than the correct value. In fact, it is absolutely impossible to choose a dimensioning value lower than the true value since there is no longer a guarantee that the loss probability is lower than . As shows Figure 3.3, it turns out that the values returned by the Gaussian method are always under the true value. Thus this annihilates any possibility to use the Gaussian approximation for dimensioning purposes.
46
3. Robust methods for LTE and WiMAX dimensioning Going one step further, according to (3.4), one may nd such that 1 Q() m(3, ) (3) Q () + E = 6 f d +
and then use 3.5 +

E
as an approximate guaranteed value of Navail . By guaranteed, we mean that according to (3.4), it holds for sure that the loss probability with this value of Navail is smaller than even if there is an approximation process during its computation. Since the error in the Edgeworth approximation is of the order of 1/, instead of 1/ for the Gaussian approximation, one may hope that this method will be ecient for smaller values of . It turns out that for the data sets we examined, E is of the order of 107 /, thus this method can be used as long as 107 / . Otherwise, as for the Gaussian case, we are reduced to nd such that 1 Q() m(3, ) (3) Q () = 6
and consider 3.5 + E f d + but we no longer have any guarantee on the validity of the value. As Figure 3.3 shows, for the considered data set, Edgeworth methods leads to an optimistic value which is once again absolutely not acceptable. One can pursue the development as in (2.25) and use (3.5), thus we have to solve 1 Q() m(3, ) (3) m(3, 1)2 (6) m(4, 1) (4) Q () Q () + Q () F 6 72 24 = .
5 function greater than 1 For the analog of 3.4 to hold, we have to nd a Cb [x, ) but smaller than 1[xlag, ) with a fth derivative smaller than 1. Looking for in the set of polynomial functions, we can nd such a function only if lag is greater than 6.5 (for smaller value of the lag, the fth derivative is not bounded by 1) thus the dimensioning value has to be chosen as: f d + . 6.5 + E
For the values we have, it turns out that F is of the order of 109 3/2 which is negligible compared to = 104 , so that we can eectively use this method for 104 . As it is shown in Figure 3.3, the values obtained with this development are very close to the true values but always greater as it is necessary for the guarantee. The procedure should thus be the following: compute the error bounds given by (3.3), (8.10) and (3.5) and nd the one which gives a value negligible with respect to the threshold , then use the corresponding dimensioning formula. If none is suitable, use a ner Edgeworth expansion or resort to the concentration inequality approach. Note that the Edgeworth method requires the computations of the rst three (or ve) moments, whose lengthiest part is to compute the k, l which is also a step required by the exact method. Thus Edgeworth methods are dramatically simpler than the exact method and may be as precise. However, both the exact and Edgeworth methods suer from the same aw: There are precise as long as the parameters, mainly and , are perfectly well estimated. The value of is often set empirically (to say the least) so that it seems important to have dimensioning values robust to some estimate errors. This is the goal of the last method we propose.
47 According to Theorem 11, if we nd such that g( and Navail =

E
L log()L2 ) = 2 2 E f d E f d f d + L2 f 2 d,
E
(3.8)
we are sure that the loss probability will fall under . However, we do not know a priori how larger this value of Navail than the true value. It turns out that the relative oversizing increases with from a few percents to 40% for the large value of and hence small values of Navail . For instance, for = 4.2, the value of Navail given by (3.8) is 40 whereas the exact value is 32 hence an oversizing of 25%. However, for = 4.12, which is 2% away from 4.2, the required number of subchannels is also 40. The oversizing is thus not as bad as it may seem since it may be viewed as a protection against trac increase, epistemic risk (model error) and estimate error.
Figure 3.3: Estimates of Navail as a function of by the dierent methods
48
49
Chapter 4
An analytic model for evaluating outage and handover probability of cellular wireless networks
Contents
4.1 4.2 4.3 4.4 4.5 4.6 Introduction System model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
Outage analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52 Handover analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . 54 Numerical results and comparison to the hexagonal model . . 55 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
4.1
Introduction
In a wireless network, nodes can be modeled by a xed or a stochastic pattern of points on the plane. Fixed points models can contain a nite or an innite number of points and usually form a lattice. This approach fails to capture the irregularity and randomness of a real network. For example, to model a wireless cellular network, the hexagonal cellular network is the most frequently used one. In reality, even if the base station (BS) nodes are xed, it is not true that they are periodically distributed. Recently, stochastic model gained much interest. Node patterns can be represented by a stochastic process on the plane such as Poisson point process. It is worth to note that stochastic models, although more complicated at rst sight, usually lead to elegant and easy calculated formulas. Actually, all the insights obtained when studying both types of models are useful for the design or the dimensioning processes of a network. In the literature, it is very often assumed that a mobile, once active in the network, is served by its nearest BS. This holds if the eect of fading is not taken into account in the propagation model. This assumption results in a so called Poisson-Voronoi cells model (for example, [8], page 63) : the domain of the plane taken in charge by a given BS is the cell of the Voronoi tessellation it is the center of. We are interested in a system which is spatially static but with some time evolutionary elements. As its name indicates, slow fading varies slowly, i.e. it may be considerered as constant on a duration of the order of a
4. An analytic model for evaluating outage and handover probability of 50 cellular wireless networks second; whereas Rayleigh fading varies much more rapidly, at the scale of a micro-second. Since we work at the time scale of a slot, i.e. of the order of micro-second, we may consider the slow fading to be constant over the period of analysis and we assume that the Rayleigh fading changes each time slot. Then, we make the very natural assumption that a mobile is served by the BS that provides it the strongest mean signal power. Mean signal power means that the eect of Rayleigh fading is averaged over a few slots. Thus, the mean signal power depends only on path loss and slow fading. Once the mobile is attributed one BS, the signal received by this BS is the useful signal, and other signals received from other BS using the same frequency are considered as interference. It is not true if we consider for example an advanced system in which the base stations are cooperative. However our model covers almost all other existing cellular networks. To model the frequency reuse, we add a label to each BS which represents its frequency band. A BS interferes only with the other BSs that have the same label. In addition to the interference, the local noise can intervene. For a mobile to communicate with a BS, the signal-to-noise-plus-interference ratio (SINR) at this mobile location must exceed some threshold, in this case the mobile is covered, otherwise it is said to be in outage. If the mobile is in outage during several consecutive time slots, a handover decision has to be made. It is thus of paramount importance to compute the outage probability and the handover probability as explicitly as possible. In [43], Haenggi showed that the path loss fading process is a Poisson point process on the real line in the case of path loss exponent model. In [19] Baccelli and al. found analytic expressions for outage probability of networks where each node tries to connect with a destination at xed distance or to the nearest node in case of Rayleigh fading. In [44], Kelif et al. found an outage probability expression for cellular network by mean of the so-called uid model. In [45], Ganti et al. developped interesting results about temporal and spatial correlation of wireless networks. In [46] and [47], outage probability of regular hexagonal cellular networks with reuse factor and adaptive beamforming were studied by simulation. This paper is organized as follows. In Section 5.2 we describe our model. In Section 4.3, we calculate the outage probability. In Section 4.4, we calculate the handover probability. Section 4.5 shows the numerical results and the dierence between our model and the traditional hexagonal model.
4.2
System model
Given a BS (base station) located at y , of transmission power P , and an MS (mobile station) located at x, the mobiles received signal has average power L(y x)P where L is the path loss function. The most used path loss function is the path loss exponent law L(z ) = K |z | where |z | refers to the Euclid norm of z . The parameter K depends on the characteristics of the antenna and the path loss exponent , typically in the range (2, 4) characterizes the environment under study. Actually, this path loss model gives nice closed formulas but is not correct for small distances as it implies an almost innite power close to the BS. It is thus often preferable to consider the modied path loss exponent model L(z ) = K (max{R0 , |z |}) where R0 is a reference distance. In addition to the deterministic large scale eect, there are two random factors that have to be considered. The rst one, called shadowing, represents the signal attenuation caused by large obstacles such as buildings. The second, called fast fading, represents the impact of multi-path. The shadowing can be considered as constant during a period of communication of a
51 mobile while the fast fading changes at each time slot. If no beamforming technique is used, the received signal power from the BS y to the MS x at the time slot l is Pyx [l] = ry,x [l]hyx L(y x)P, where {hyx }x,yR2 are copies of a random variable H while {ryx [l]} are independent copies of R which is an exponential random variable of mean 1/. We suppose that for each x, the random variables (hyx , y R2 ) are independent, and pH (resp. FH ) denotes their PDF (resp. complementary CDF). The most used shadowing random model is log-normal shadowing, for which H is a log-normal random variable. In this case, we can write H 10G/10 where G N (0, 2 ). We now consider the conventional beamforming technique with nt antennas. The power radiation pattern for a conventional beam-former is the product of the array factor times the radiation pattern of a single antenna. If is the direction towards which the beam is steered, the array gain in the direction is given by ([47],[47]): sin2 (nt 2 (sin( ) sin()) g( ), 2 nt sin ( 2 (sin( ) sin()) where g( ) is the gain in the direction with one antenna. For simplicity we assume that the BS always steers to the direction of the served MS and the gain g( ) is positive constant on (/2, /2) and 0 otherwise (zero front-to-back power ratio). Hence, the interference signal power from a BS to an MS attached to another BS using the same frequency, in the direction , will be reduced by a factor of: a( ) = 1{(/2,/2)} sin2 (nt 2 (sin( ))) 2 2 nt sin ( 2 (sin( )))
If the beamforming technique is not used a( ) = 1. We assume that the bandwidth is split in k non interfering sub-bands. Thus, for a mobile at position x, any BS is characterized by three quantities: y its position, e the sub-band in which it operates and 1 = hxy L(y x)P . Once being in the network, the mobile x is attached to (or served by) the BS that provides the best average signal strength in time : it is attached to the BS which has the minimal . We denote by y0 the position of the chosen BS and by (yn , n 1) the positions of the other BSs. Sub-bands and s are indexed accordingly. Assume that each BS using frequency e0 is always serving an MS, and denote by i the argument of the segment [x, yi ]. The SINR at time slot l is given by: sx [l] = N+
1 ry0 x [l]0 1 i=0 1{ei =e0 } a(i )ryi x [l]i
(4.1)
1 where N is the noise power, assumed to be constant. The term Ix = i=b(x) 1{ei =e0 } a(i )ryi x [l]i is the sum of all interference. In order to communicate with the attached BS, the SINR must not fall below some threshold T . We assume that the base stations are distributed in the plane according to a Poisson point process B of intensity B for any details on Poisson point process we refer to [8]. The frequency ei at which operates yi is chosen uniformly in {1, , k} where k is the frequency reuse factor. The BSs that have the same mark interfere between themselves. Our reuse model can be considered as a worst case scenario since the sub-bands are distributed at random, in contrast with planned network patterns where frequencies are attributed to BSs in order to minimize interference. The subsequent computations rely mainly on the following theorem.
4. An analytic model for evaluating outage and handover probability of 52 cellular wireless networks Theorem 15. The family of random variables = {(hy,x L(y x)P )1 , y B } is a Poisson point process on R+ with intensity d(t) = B B (t)dt where B ( ) = R2 FH ((L(z )P )1 ) dz. Proof. Dene the marked point process x = {yi , hyi x } i=0 . It is a Poisson point process of intensity B dy fH (t)dt because the marks are i.i.d. Consider the probability kernel p((z, t), A) = 1{(L(z )P t)1 A} for all Borel A R+ and apply the displacement theorem [8, Theorem 1.3.9], to obtain that is a Poisson point process whose intensity measure we denote by . Moreover, for any ([0, ]) = B = B This concludes the proof. By straightforward quadatures, we get the following proposition. Proposition 16. If L(z ) = K (max{R0 , |z |}) then: B ( ) = C1
2 2
R2 R R2
1{t(P.L(z )P )1 } pH (t)dzdt
FH ((L(z )P )1 )dz = B B ( ).
R 0 P K
t pH (t)dt,
(4.2)
where C1 = (P K ) . For lognormal shadowing H 10G/10 where G N (0, 2 ) and we have: 2 2 ln ln(P KR0 ) 21 ( 1 )2 Q( ) (4.3) B ( ) = C1 e 1 where Q(u) =
1 2 u2 /2 du u e
and 1 =
ln 10 10 .
For the exponent pathloss model, it is sucient to put R0 = 0 in the above formulas. This particular result could be derived from [43]. We observe that the distribution of the 2 point process does depend only on E (H ) but not on the distribution of H itself. This phenomenon can be explained as in [48, Page 159].
4.3
Outage analysis
The mobile at x suers an outage at time slot l whenever its SINR falls below a threshold T at this slot. For the sake of notations,in this Section, we drop the index l as it is xed. Theorem 17. The outage probability is given by po := P (sx < T ) = 1 B where D ( ) =
0
B ( )eB B ( )N T 2k D( ) d
(4.4)
B ( )(1
+ /T a( ))1 d .
Proof. Since ry0 x is an exponential r.v. of mean 1/ we have: P (sx T |0 = ) = P (ry0 x T (N + Ix ( )) | 0 = ) = eN T LIx ( ) (T ) = E (eT (N +Ix ( )) | 0 = )
53 where Ix ( ) is the distribution of the random variable Ix given (0 = )) and LIx ( ) is its Laplace transform. Given (0 = ), according to strong Markov property, the point process {i }i>0 is a Poisson point process on (, ) with intensity B B ( )d . By thinning, the point process {i }{i>0,ei =e0 } is a Poisson point process on (, ) with intensity k1 B B ( )d . Hence, LIx ( ) can be calculated as follows (see [8]): = e
B a( )u1 R ))d 2k B ( )(1E (e B
LIx ( ) (u) = e 2k
B ( )d
dr
er (1ea()ur
)d
= exp Thus, we get
B 2k
B ( )
d 1 + (ua( ))1
B
= P (sx T |0 = ) = eN T 2k D( ) .
(4.5)
Since the distribution density of 0 is B B ( )eB B ( ) , by averaging over all 0 we obtain (4.4). Proposition 18. In the interference-limited regime (N = 0), we have po (T ) = 1 B If L(z ) = K |z | we have: po (T ) = 1 where M := M (k, T, ) = 1 +
1 2k 0 du 1 1+(T.a( ))1 u 2 0
B ( )eB B ( ) 2k D( ) d.
(4.6)
eM G 2 d
(4.7) and G = N T (B C ) 2 . If
L(z ) = K |z | and N = 0 we have:
po (T ) = 1 as sx [l] = N +
1 M
(4.8)
Some interesting facts are observed from these results: Rewrite the expression of SINR
1 ry0 x [l]0 1 i=0 1{ei =e0 } a(i )r yi x [l]i
where r y0 x [l] = ryi x [l]. Since ryx [l] is an exponential random variable of mean 1/, r y0 x [l] is an exponential random variable of mean 1. Hence by the above equation it is expected that the outage probability depends on the product N but not directly on and N . It is an increasing function of N which is conrmed by (4.4). The fact that the outage probability is an increasing function of and N is quite natural, increasing of noise or the fast fading inuence always deteriorate the system performances. It is also expected that in the interference limited case (N = 0) the outage probability does not depend on . It is conrmed by (4.6). Physically it means that in the absence of noise, the fast fading modies the channels (from the MS to each BS) characteristics by the same factor, thus the SINR does not change.
4. An analytic model for evaluating outage and handover probability of 54 cellular wireless networks In the interference limited scenario with the exponent pathloss model, the outage probability does not depend neither on , nor on the BS density B and nor on the distribution of shadowing H . It is due to the scaling properties of the pathloss function and of the Poisson point process. The outage probability is a decreasing function of the pathloss exponent , reecting the fact that bad propagation environment deteriorates the received SINR. In the presence of noise N > 0 and still for the exponent pathloss model case, the outage probability is an increasing function of B . Hence, it can be thought that the more an operator installs BSs, the better the network is. In addition, if the density of BSs goes to innite then outage will never occur. However it is not true. In fact, if the density of BSs is very high, the distances between a MS and BSs tend to be relatively small. Hence, the exponent pathloss model is no longer valid since it is not accurate at small distances. If the modied exponent pathloss is used, the outage probability must converge to 0. The 2 outage probability is also an increasing function of E (H ), and if the shadowing H follows lognormal distribution then the outage probability will be an increasing function of . We recover an other well known fact: the increasing of uncertainty of the radio channel deteriorates the performance of the network.
4.4
Handover analysis
If the MS is in outage for n consecutive time slots, a handover decision has to be made. Keep in mind that only the Rayleigh fast fading changes each time slot. Let Al be the event that the mobile is in outage in the time slot l, and Ac l its complement and observe m c m c n that in fact P (i=1 Aji ) = P (i=1 Ai ). By denition pho := P (n i=1 Al+i1 ) = P (i=1 Ai ). We have
n
pho = 1 +
m=1 n
(1)m (1)m
j1 =...=jm {1,..,n}
c P (m i=1 Aji )
= 1+
m=1
n! c P (m i=1 Ai ) m!(n m)! n! qm , m!(n m)!

B
Theorem 19. The handover probability is given by:

n
pho = 1 +
m=1
(1)m
where qm =
c P (m i=1 Ai )
is given by:
qm = and Dm ( ) =

B B ( )eB B ( )N T 2k Dm ( ) d,
1 )m )d . ( 1+T a ( ) 1
0 B ( )(1
c Proof. We need to calculate the probability P (m i=1 Ai ) that is the probability that the mobile is covered in m dierent time slots.
c P (m i=1 Ai |0
= E (e(mT N + =e
mN T
= P (ry0 x [i] (T N + Ix [i]), i = 1 . . . m|0 = )

m i=1 Ix ( )[i])
= P (sx [1] T, ..., sx [m] T |0 = ) |0 = )
m i=1 Ix ( )[i]
(T )
55 where Ix ( )[i] is the distribution of the random variable Ix [i] given (0 = ). We have :
m
Ix ( )[i] =
i=1
j =1
m 1 1{ei =e0 } i a(i )( i=1
ryi x [i])
As the random variables ryi x [i] are independent copies of the exponential random variable R, the random variables m i=1 ryi x [i] are also i.i.d and the common Laplace transform of the latter is : m L m (u) = (LR (u))m = ( ) i=1 ryi x [i] +u The Laplace transform of L
m i=1 Ix ( )[i]
m i=1 Ix ( )[i]
is now:

(u) = e
B 2k
B ( )(1(
)m )d +a( )1 u
Proceeding as for Theorem 17, we get qm =

0
B B ( )eB B ( )N T 2k Dm ( ) d
The result follows. We can obtain more closed expression for qm in some special cases. Proposition 20. In the interference limited regime N = 0, we have: qm =
0
B B ( )eB B ( ) 2k Dm ( ) d
0
If L(z ) = K |z | then: qm = where Mm = 1 + 1 2k
eMm G 2 d
d
1
(1 (
1 1 + T a( )u 2
)m )du.
If N = 0 and L(z ) = K |z | we have: qm = 1/Mm . From these computations, the same kind of conclusions as for outage probability can be drawn.
4.5
Numerical results and comparison to the hexagonal model
We place a MS at the origin o and consider a region B (o, Rg ) where Rg = 10, 000(m). The BSs are distributed according to a Poisson point process in this region. The path loss exponent model is considered. The default values of the model parameters are K = 20 dB, P = 0 dB, nt = 8 and = 1. In the literature, the hexagonal model is widely used and studied so we would like to compare two models. For a fair comparison, the density of BSs must be chosen to be the same, i.e the area of an hexagonal cell must be 1/B . Unlike the Poisson model where each BS is randomly assigned a frequency, in the hexagonal model, the frequencies are well assigned so that an interfering BS is far from the transmitting BS and BSs of dierent
4. An analytic model for evaluating outage and handover probability of 56 cellular wireless networks
= 1.9099.106, = 6, = 1, N = 174(dBm), k = 7 0.35 = 4, Poissonsimulation = 4, Poissonanalytic = 4, Hexagonalsimulation = 3, Poissonsimulation = 3, Poissonanalytic = 3, Hexagonalsimulation
0.3
0.25 Outage probability
0.2
0.15
0.1
0.05
10 T (dB)
12
14
16
18
20
Figure 4.1: Outage probability vs SINR threshold
frequency are grouped in reuse patterns. The reuse factor k in the hexagonal model is determined by k = i2 + j 2 + ij where integers i, j are the relative location of co-channel cell. Figure 4.1 shows the outage probability versus the SINR threshold of the Poisson model and the hexagonal model in the case k = 7. As we can see, the outage probability in the case of Poisson model is always greater than that of hexagonal model as expected. The dierence is about 8 (dB) in the case = 4 and 6(dB) in the case = 3. In Figure 4.3, we can see that the outage probability is a decreasing function of as theoretically observed. In Figure 4.4, we see if the reuse factor k increases, the MS has to do less handover. Thus, increasing the reuse factor has a positive eect on the system performance not only in term of outage but also in term of handover.
4.6
Conclusion
In this paper we have investigated the outage and handover probabilities of wireless cellular networks taking into account the reuse factor, the beamforming, the path loss, the slow fading and the fast fading. We valid our model by simulation and compare numerical results to that of hexagonal model. The analytic expressions derived in the this paper can be considered as an upper bound for a real system.
57
N = 174 (dBm), k = 7, = 6 (dB), = 1 0.1 0.09 0.08 0.07 Handover probability 0.06 0.05 0.04 0.03 0.02 0.01 0 Hexagonal model, = 3, n = 6 Hexagonal model, = 3, n = 3 Poisson model, = 3, n = 3 Poisson model, = 3, n = 6 Hexagonal model, = 4, n = 6 Hexagonal model, = 4, n = 3 Poisson model, = 4, n = 3 Poisson model, = 4, n = 6
10 T (dB)
12
14
16
18
20
Figure 4.2: Handover probability vs SINR threshold
N = 174 (dBm), T = 5 (dB), = 6(dB), = 1 0.1 0.09 0.08 0.07 Outage probability 0.06 0.05 0.04 0.03 0.02 0.01 0 2.4 k=7 k=3 k = 12
2.6
2.8
3.2
3.4
3.6
3.8
Figure 4.3: Outage probability vs path loss exponent , Poisson model
4. An analytic model for evaluating outage and handover probability of 58 cellular wireless networks
N = 174 (dBm), T = 5 (dB), = 6(dB), = 1 0.025 k=3 k=7 k = 12 0.02
Handover probability
0.015
0.01
0.005
0 2.4
2.6
2.8
3.2
3.4
3.6
3.8
Figure 4.4: Handover probability vs path loss exponent , Poisson model, n = 3
59
Chapter 5
On noise-limited networks
Contents
5.1 5.2 5.3 5.4 5.5 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59 Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60 Poisson point process of path loss fading . . . . . . . . . . . . . 61 Capacity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65 Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70 5.5.1 5.5.2 5.5.3 5.5.4 5.5.5 5.6 Number of users in a cell . . . . . . . . . . . . . . . . . . . . . . Number of users in outage in a cell . . . . . . . . . . . . . . . . . Number of covered users in a cell . . . . . . . . . . . . . . . . . . Total bit rate of a cell . . . . . . . . . . . . . . . . . . . . . . . . Discussion on the distribution of So (f ) . . . . . . . . . . . . . . . 70 71 71 71 72
Conclusion
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
5.1
Introduction
Cellular network is a kind of radio network consisting of a number of xed access points known as base stations and a large number of users (or mobiles). Each base station covers a geometrical region known as a cell and serve all users in this cell. Interference and noise are two factors annoying communications in cellular wireless networks. Noise is unavoidable and comes from natural sources. Interference come from users and base stations. The use of recent technologies such as SDMA (spatial division multiple access) and MIMO (multiple input multiple output) can reduce signicantly interference so that we can hope that in a near future the impact of interference will be negligible and noise will become the only factor harming the network. The best case is when interference from other cells are perfectly canceled, the network is then said to be in noise limited regime. We introduce here a framework to study this kind of network. In the existing literature, base stations (BS) locations are usually modeled as an ideal regular hexagonal lattice. In reality, base stations are irregularly located, especially in an urban area, and the cell radius is not the same for each BS. In this chapter, we model the base station locations as an homogenous Poisson point process B of intensity B . Such a model comes from stochastic geometry. It is suciently versatile by changing B to cover a wide number of real situations and it is mathematically tractable. For an introduction
60
5. On noise-limited networks
to the usage of stochastic geometry for wireless networks performances, we refer to [49]. Theory and number of pertinent examples can be found in [8] and [50]. For all theoretical details, we refer to the rst opus. To model cellular network cells, Voronoi tessellations are frequently used. It is based on the assumption that each user is served by the closest BS. Unfortunately, this is not always very accurate since in real life, a mobile connects to the best BS it can have, i.e., the BS which oers it the best Signal over Noise Ratio. The best BS is not always the closest because of the fading environment. In this chapter, we analyze the impact of fading by considering that users are served by the base station providing the best signal power. The location of users in the plane are modeled as another homogenous Poisson point process M of intensity M . While cellular networks like GSM and GPRS provided only voice service and low data transmission rate, recent and emergent wireless cellular networks such as WIMAX or LTE oer higher data rate and other services requiring high throughput such as video calls. Each service requires a dierent level of signal to noise ratio (SNR). If the SNR does not reach a required threshold due to the radio condition, the service cannot be established or may be interrupted. Such calls are said to be in outage. The outage probability is one of the key measurement of the network performance. We aim to determine the outage probability of noise limited network, or equivalently the distribution of SNR, which turns out to be equivalent to determine the distribution of the smallest path loss fading. In fact, there have been some works dealing with the outage probability of noise limited wireless network, but almost all of them consider the exponent path loss model. We here derive a formula for outage probability taking into account a general model of path loss. Once the distribution of SNR of a user is determined, the distribution of functionals related to SNR can be easily derived. In some situations, we have to study the distribution of the sum of a functional for all users in a cell. For example, in an OFDMA noise limited cellular system, the number of sub channels required for a user demanding a particular service depends on its SNR. If the total number of sub channels of all users in a cell excesses the number of available sub channels in this cell then at least one user is blocked. The probability of that to happen, sometimes called unfeasibility probability, contains extremely important information on the performance of the network. Since it is often impossible to nd the explicit probability distribution of additive functionals, we calculate the expectation, and bounds on the variance of such random variables.
5.2
Model
Consider a BS (base station) located at y with transmission power P and a mobile located at x. The mobiles received signal has average power L(y x)P where L is the path loss function. The most used path loss function is the so-called path loss exponent model L(z ) = K |z | , where |z | refers to the Euclidean norm of z . This function gives raise to nice closed formulas but is rather unrealistic: Close to the BS, the signal is innitely amplied. A more realistic model is the modied path loss model given by:
L(z ) = K min{R0 , |z | }
where R0 is a reference distance and K a constant depending on the environment. In addition to this deterministic large scale eect, we consider the fading eect, which is by
61 essence random. The received signal power from a BS located at y to a mobile unit (MU for short) located at x is given by Pyx = hy, x L(y x)P, where {hy, x }x,yR2 are independent copies of a random variable H . Most used fading random models are log-normal shadowing and Rayleigh fading. The log-normal shadowing is such that H is a log-normal random variable and we can write H 10G/10 where G N (0, 2 ). The Rayleigh fading is such that H is an exponential random variable of parameter . We can also consider the Rayleigh-Lognormal composite fading, in this case the fading is the product of the log-normal shadowing factor and the Rayleigh fading factor. It is worth noting that the log-normal shadowing usually improves the network performance while Rayleigh fading usually degrades performances. We assume that once in the network, a mobile is attached to the BS that provides it the best signal strength. If the power received at this point is greater than some threshold T , we say that x is covered. If x is not covered by any BS then a MU at x cannot establish a communication and thus is said to be in outage. In the case of path loss exponent model with no fading (H = constant), the best BS for given mobile is always its nearest BS. We assume that the point process of BSs B = {y0 , y1 , ...} is an homogenous Poisson point process of intensity B on R2 and that users are distributed in the plane as a Poisson point process M = {x0 , x1 , ...} of intensity M . To avoid any technical diculty, from now on, we make the following assumptions: Assumption 1. Assume that: 1. All random variables hyx (x, y R2 ) are independent. 2. H admits a probability density function pH . Its complementary cumulative distributive function is denoted by FH , i.e., FH ( ) = P (H ) = 3. Dene B ( ) =
R2
pH (t) dt > 0.
FH ((L(z ) )1 ) dz . Then, we have 0 < B ( ) < for all > 0.
5.3
Poisson point process of path loss fading
For each location x on R2 , consider the path loss fading process Shx = {syx := hyx L(y x))1 , y B }. The next proposition follows from [43]. Proposition 21. For any x, Shx is a Poisson point process on R+ with intensity density d(t) = B B (t)dt. In addition, B (0) = 0 and B () = . For any point x, we can reorder the points of Shx . We denote ordered atoms of Shx by x < x < . . .. The CDF and PDF of x are easily derived according to the property 0 0 m 1 of Poisson point processes:
x is given by: Corollary 1. The complementary cumulative distribution function of m m x P (m
> t) = e
B B (t) i=0
(B B (t))i , i!
62
1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Figure 5.1: Triangles represent BS, plus represent MU. Dotted polygons are Voronoi cells induced by BS. A line between a BS and an MU means that the BS serves the MU. A mobile may be not served by the BS closest to it, due to fading.
63 and its probability density function is given by

x (t) = pm
+1 B (t)(B (t))m B B (t) m B e m!
(5.1)
x > t) is equivalent to the event (in the interval [0, t], there are at Proof. The event (m most m points) and the number of points in this interval follows a Poisson distribution of mean B B (t). Thus, we have: m x P (m
> t) = e
B B (t) i=0
(B B (t))i i!
The PDF is thus given by

x (t) pm
= =
x P (m > t) t B B (t)eB B (t)

m
+ =
B B (t)eB B (t) e
1 m+1 B B (t)(B (t))m B B (t)
(B B (t))i1 (B B (t))i (i 1)! i!
m!
The proof is thus complete. Corollary 2. If L(z ) = K |z | then: B ( ) = C. , where C = K E (H ). Proof. The path loss function depends only on the distance from the BS to the user. By the change of variable r = |z | and by integration by substitution, we have: B ( ) = 2
0 0 0
2 2 2 2
r 1{tK r } pH (t) dr dt
(tK )1/
= 2
pH (t) dt
2
r dr
0
2
= (K )
0
2
pH (t)t dt
2
= (K ) E (H ) Hence the result. We observe that the distribution of the point process Shx does depend only on E (H ) but not on the distribution of fading H itself. This phenomenon can be explained as in [48](page 159). If the fading is log-normal shadowing, i.e H 10G/10 where G N (0, 2 ) then E (H ) = e 2 where 1 =
2 2 2 21 2
then E (H ) = + 1, 0) where (a, b) = b ta1 et dt is the upper incomplete gamma function. Similarly to the distance to m-th nearest BS (which can be found in [51]), the distrix can be characterized as follows: bution of m-th less strong path loss fading m
2 (
ln(10) 10 .
If the fading is Rayleigh fading, i.e H exp()
64
x is distributed according to the generalized Gamma Corollary 3. If L(z ) = K |z | , m distribution:

2 2 eB Ct x (t) = (B C )m+1 t (m+1) pm m! 2
Proof. This is a consequence of Proposition 1 and Lemma 2. We can also investigate more general and realistic path loss model.
Corollary 4. If L(z ) = K min{R0 , |z | } then:
B ( ) = C1
2
R 0 K
t pH (t) dt,
(5.2)
where C1 = K . In addition, we have: 2 2 B ( ) = 1 B ( ) + R0 pH

R0 K
(5.3)
If the fading is lognormal shadowing H 10G/10 where G N (0, 2 ) then we have: B ( ) = C1 e( where Q(a) = 1 2 H exp() then
u2 /2 a e
2 21 2 )
ln ln(KR0 ) 21 1 ln 10 10 .
du is the Q-function and 1 =

2
If the fading is Rayleigh
B ( ) = C1
1+
2 R0 , K
Proof. Similarly to the path loss exponent model case, we have: B ( ) = 2

0 R0
rFH ((max{R0 , r }) (K )1 ) dr
rFH (R0 (K )1 ) dr + 2
R0 K
= 2
0
R0
rFH (R0 (K )1 ) dr (tK )1/
2 = R0 FH (R0 (K )1 ) + 2
2
pH (t) dt
R0
r dr
= C1
R 0 K
t pH (t) dt
We then obtain Equation (5.2). Now dierentiate the two sides of that equation to get: B ( ) = =
2 2 C1 1
R0 K
t pH (t) dt +
R0 K
2 C 1 R0
K .
pH
R0 K
2 1 2 B ( ) + R0 pH
65 That yields Equation (5.3). In the case of lognormal shadowing we have: B ( ) = C1

2
R0 K
1
2t 21
t e
) (ln t2 21
dt du
2 2 (u 1 )2 2 21
= C1
ln
R0 K
21
ln
R0 K
e e 2 1
2u
u2 2 21
= C1 e = C1 e
2
2 ( 1 )2
2 21
du
21 2 )
ln ln(KR0 ) 21 1
In the case of Rayleigh fading we have: B ( ) = C1 = C1

2
R 0 K 2
t et dt 1+
2 R0 , K
by a change of variables. Hence the result. Corollary 5. Let T be the value of attenuation above which a communication is not feasible. The number of BS covering a point x is distributed according to the Poisson distribution of parameter B B (T ). In particular, the outage probability is given by
x P (0 > T ) = eB B (T ) .
Proof. The path loss fading Shx is a Poisson point process on R+ with intensity B B (t) dt, so the number of point on the interval (0, T ) is distributed according to Poisson distribution of parameter B B (T ). Figure 5.2 represents the outage probability for dierent models of fading. This shows that the curves of modied path loss exponent model is generally higher than those of path loss exponent model but they are very close in the low outage region.
5.4
Capacity
In this section, we calculate the mean of any capacity function of a user. Remark that since the system is spatially stationary the statistic of the path loss fading and the capacity of a user does not depend on his position. Since the PDF and the CDF of the path loss x have been already calculated in Proposition 1, the mean of a capacity function fading 0 of a user follows immediately. In particular: Theorem 22. The average capacity per user is
x E (f (0 )) = B 0
B ( )eB B ( ) f ( ) d
(5.4)
66
1 Shadowing, L(z) = K|z| 0.9 0.8 0.7 Outage probability 0.6 0.5 0.4 0.3 0.2 0.1 0

Rayleigh fading, L(z) = K|z|
Rayleigh fading, L(z)=K min{R ,|z| } Shadowing, L(z)=K
0 min{R ,|z| } 0
5 T
10 x 10
6
Figure 5.2: Comparison of outage probability between propagation models. For lognormal shadowing = 4(dB ), for Rayleigh fading = 1; K = 102 , = 2.8.
In the case of path loss exponent model L(z ) = K |z | , we have:

x E (f (0 )) = Lf (B C )
(5.5)
where Lg (s) = f (t) = f (t 2 ).
st g(t) 0 e
dt is the Laplace transform of the capacity function g and
Proof. Equation (5.4) comes from Proposition 1. If the path loss exponent model is considered, then we have:
x E (f (0 )) = B 0
2 2 2C 1 B C e f ( ) d
=
0
2 ) d1 eB C1 f (1
by the change of variable 1 = 2 . The statistic of the cell capacity So (f ) is more dicult to analyze. In this section, we calculate its mean m(f ) and lower bound and upper bound of its variance v (f ). We state the following lemma, which is straightforward due to Assumption 1 but still useful: Lemma 23. Given a xed conguration B of BSs, the Poisson point processes of path loss fading Shx and Shy are independent for any two dierent points x, y . Lemma 24. Let z = y x. The PDF of syx is given by psyx (t) = 1 pH l(z )t2 1 L(z )t .
67 Proof. We have P (syx < t) = P (hyx > = FH The density probability function is then psyx (t) = 1 pH L(z )t2 1 L(z )t . 1 ) L(z )t .
1 L(z )t
Theorem 25. The expectation of the cell capacity of the typical BS is m(f ) = M
0
B ( )eB B ( ) f ( ) d
(5.6)
In the case of path loss exponent model L(z ) = K |z | , we have: m(f ) = M L (B C ) B f (5.7)
x )f (s ) Proof. Given a xed conguration of BSs B , the random variables 1(sox < 0 ox M = obtained from all x R2 are independent. Thus, the marked point process xi (xi , 1(soxi < 0 )f (soxi )) is a Poisson point process. Using the Campbell theorem we have: x E (So (f ) | B ) = M E (1(sox < 0 )f (sox ) | B ) dx R2
As a consequence, E (So (f )) = E M = M
x E (1(sox < 0 )f (sox ) | B ) dx
R2
R2
x E (1(sox < 0 )f (sox )) dx
In virtue of Lemma 24, Proposition 21 and Collary 1, we have: E (So (f )) = M = M = M = M

0 x E (1(sox < 0 )f (sox )) dx R2 0 0 x psox (t)P (t < 0 )f (t) dt dx
R2
f (t)eB B (t) dt
R2
FH t
f (t)eB B (t) dt
1 dx L(x)t 1 F( ) dx L(x)t R2
= M
0
f (t)eB B (t) B (t) dt
For the case of path loss exponent model, Equation 5.7 follows easily. This completes the proof.
68
Equation (5.6) has the following interpretation: the mean cell capacity is the product of the mean number of users per cell and the mean capacity per user. Theorem 26. Given two capacity functions f, g we have : cov(So (f ), So (g)) m(f.g) In particular, var(So (f )) m(f 2 ).
x ) and f (0) = 0, g (0) = 0, we have : Proof. For simplicity, let x = sox 1(sox < 0
(5.8)
cov (So (f )So (g)) = E (cov (So (f ), So (g) | B )) = T1 + T2 T3
+ E (E (So (f ) | B )E (So (g) | B )) E (So (f ))E (So (g))
It is clear that T3 = m(f )m(g) Consider the rst term. Remind that we have assumed that all random fading {hyx }y,xR2 are independent, so given a xed conguration B of BSs, the random variables {x }xR2 are independent. Hence by Campbell formula we have: T1 = M E = M = M
R2
E (f (x )g(x ) | B ) dx
R2
E (E (f (x )g(x ) | B )) dx E (f (x )g(x )) dx
R2
= m(f.g) Now consider the second term T2 = 2 ME E (f (x ) | B ) dx

R2 R2
R2
R2
E (g(x ) | B ) dx
= 2 ME = 2 M = 2 M = 2 M
E (f (x ) | B )E (g(y ) | B ) dx dy
R2
R2
E (E (f (x ) | B )E (g(y ) | B )) dx dy E (f (x ))E (g(y )) dx dy

0 y x P (sox < 0 , soy < 0 | sox = t1 , soy = t2 )
R2
R2
R2
R2
f (t1 )g(t2 )psox (t1 )psoy (t2 ) dt1 dt2 dx dy by remarking that x and y are independent if x = y (Lemma 23). We will prove that if x = y:
y x P (sox < 0 , soy < 0 | sox = t1 , soy = t2 )
y x P (sox < 0 | sox = t1 )P (soy < 0 | soy = t2 )
69 Consider the marked point process x,y B = {yi , hyi x , hyi y )}. Since the marks are independent, it is a Poisson point process on R4 with intensity
x,u1 ,u2 ma ( dy, du1 , du2 ) = M dy pH (u1 ) du1 pH (u2 ) du2
Consider two sets

1 A1 = {(y, u1 , u2 ) : L(y )u1 t 1 }
and
1 A2 = {(y, u1 , u2 ) : L(y )u2 t 2 },
we have:
y x , soy < 0 | sox = t1 , soy = t2 ) = P (x,y P (sox < 0 B (A1 A2 ) = )
= ema
x,y
(A1 A2 )
y x | sox = t1 )P (soy < 0 = P (sox < 0 | soy = t2 )
x,y = P (x,y B (A1 ) = )P (B (A2 ) = )
ema
x,y
(A1 )mx,y a (A2 )
Thus, T2 2 M
R2 R2 0 0 y x P (sox < 0 | sox = t1 )P (soy < 0 | soy = t2 )
= m(f )m(g) The result follows.
f (t1 )g(t2 )psox (t1 )psoy (t2 ) dt1 dt2 dx dy
Theorem 27. For f and g two capacity functions, we have: cov(So (f ), So (g)) m(f.g) + m(f )n(g) m(f )m(g) where n(f ) = M
0
B (t)f (t) dt
(5.9)
Proof. We continue the proof of Theorem 26, we have to prove that T2 m(f )n(g) Indeed,
y x x P (sox < 0 , soy < 0 | sox = t1 , soy = t2 ) P (sox < 0 | sox = t1 ),
thus, T2 2 M
R2 R2 0 0 x P (sox < 0 | sox = t1 )
f (t1 )g(t2 )psox (t1 )psoy (t2 ) dt1 dt2 dx dy 1 FH ( ) dx dt2 g(t2 ) m(f ) L(x)t R2 t 0 = m(f )n(g) Hence the result.
70
600
500
400
300
200
100
0 10
10
20
30
40
50
60
70
80
Figure 5.3: Histogram of no
5.5
5.5.1
Examples
Number of users in a cell
For f0 (t) = 1, the random variable no := S (f0 ) = i=0 1(xi Co ) represents the number of users who view o as the best server, and thus will be served by o.
0
E (no ) = M = M B
B ( )eB B ( ) d
The mean number of users served by a BS is the formula (5.6) by E (So (f )) =
M B
which is easily interpreted. We rewrite
M x E (f (0 )). B
Again this is easily interpreted. The average sum rate is the product of the average user per cell and the average per user. Now apply Theorem 26, we get that M B
0 B (t)
var(no ) m(1) = We cannot apply Theorem 27 because n(f0 ) = M
dt = .
71
5.5.2
Number of users in outage in a cell
Consider f1 (t) = 1(t > T ), then So (f1 ) is the number of users in outage in the typical cell. We have m(f1 ) = M = and v (f1 ) M B B (T ) e B
T B (t) T
B ( )eB B ( ) d
M B B (T ) e , B
Note that again, we cannot apply Theorem 27 as n(f1 ) = M
dt is innite.
5.5.3
Number of covered users in a cell
Consider f2 (t) = 1(t T ), then S (f2 ) represents the number of covered users in the typical cell. We have:
T
m(f2 ) = M
0
B ( )eB B ( ) d
M = 1 eB B (T ) , B M B 1 eB B (T ) ,
v (f2 ) and v (f2 )
M B B (T ) 2 e + M eB B (T ) B B (T ) 1 + eB B (T ) B 2 B
5.5.4
Total bit rate of a cell
n We now consider the piecewise constant function f3 (t) = 1 1(Ti t < Ti+1 )ci with 0 < T1 < T2 < ... < Tn < Tn+1 and Tn+1 can be innite. If f3 is the function that represents the actual bit rate then So (f3 ) represents the total bit rates of all users in the cell. We have:
m(f3 ) = M
0 n
B ( )eB B ( )
1
1(Ti < Ti+1 )ci d
= M
i=1
ci eB B (Ti ) eB B (Ti+1 )
n
v (f3 ) M
i=1
B B (Ti ) eB B (Ti+1 ) c2 i e
72
1 Tail distribution of S (f )
o 3
0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0
Upper bound Gaussian approximation with lower bound of variance Gaussian approximation with upper bound of variance
40
50
60
70
80
90
100
110
120
Figure 5.4: Tail distribution of So (f3 )
v (f3 ) M +
i=1
B B (Ti ) eB B (Ti+1 ) + c2 i e 2 n i=1
M B
n
eB B (Ti ) eB B (Ti+1 )
i=1
B B (Ti+1 ) B B (Ti ) eB B (Ti ) + eB B (Ti+1 )
5.5.5
Discussion on the distribution of So (f )
The distribution of So (f ) does not behave like a Gaussian distribution even in the limit regimes. Take, for example, the histogram of no = So (f0 ) and that of So (f3 ) which are shown in gures 5.3 and 5.5 respectively. For the case of no fading, H = constant, in [52] the author found some approximate but not reliable bounds of the distribution of So (f ) for equivariant functions f but no approximation or bounds is found for general capacity functions. In addition, no closed expression is found for the Laplace transform of functional So (f ). In our case where the fading is considered, this is expected to be more challenging. We can nd an upper bound for the tail distribution by Chebyshevs inequality: P (So (f ) > m(f ) + t) v (f ) v (f ) + t2 m(f 2 ) + m(f )n(f ) (m(f ))2 m(f 2 ) + m(f )n(f ) (m(f ))2 + t2
73
2500
2000
1500
1000
500
0 20
20
40
60
80
100
120
140
160
Figure 5.5: A typical histogram of So (f3 )
The above inequality provides a robust upper bound for the tail distribution and valid for all capacity function f . However the gap is large (Figure 5.4). It is well known that other types of concentration inequality based on Cherno bound can give better bound. In this direction, [16], [15] and [14] provide concentration inequalities that apply for functional related to one PPP. These inequalities cannot be directly applied in our case because our target is a functional related to two independent PPPs. Actually we can combine the two independent PPPs into one united PPP by the independent marking theorem. Unfortunately the functional So (f ) of the united PPP does not satisfy the required conditions for the concentration inequalities neither on [16], [14] nor on [15]. But we believe that similar techniques used in these references can be used to derive a upper bound the tail distribution of So (f ).
5.6
Conclusion
In this chapter we introduce a general model to evaluate the outage probability and the capacity of wireless noise limited network. It is in fact an extension of models introduced in series of papers [17], [18], [52]. The main dierence is that we take into account the eect of fading, and that we assume that a user connects to the BS with strongest signal rather than the closest one. We rst show that for a particular user, the path loss fading process from all BSs seen from this user is a Poisson point process in the positive half line. We nd explicit expression for the outage probability, the expectation of capacity of a user, and the expectation of the cell capacity of the typical BS So (f ). We nd the lower bound and upper bound for the variance of the cell capacity. We consider general model for path loss and fading. The results presented in this chapter actually generalizes the results on [52]. Possible further research is to nd a way to compute the distribution of So (f ).
74
75
Part III
Energy consumption models
77
Chapter 6
Generality and basic model

Contents
6.1 6.2 6.3 6.4 6.5 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77 Model for energy consumption . . . . . . . . . . . . . . . . . . . 77 Model for mobility of users . . . . . . . . . . . . . . . . . . . . . 80 Basic model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
6.1
Introduction
In this chapter we introduce an energy consumption model for cellular network. In Section 6.2, we present the power consumption model and the energy consumption model. In Section 6.3, we present the mobility model for users. In Section 6.4, we establish the relationship between the consumed energy and the parameters of the system such as intensity of users or cell radius.
6.2
Model for energy consumption
We suppose that there is a cellular network with multiple base stations on Rd and there is a base station located at the origin o administering a geographical region C around o. We assume that there exists 0 < R1 < R such that B(o, R1 ) C B(o, R) and C is convex and compact. We dene Rinf = inf R {C B(o, R)}. For a given spatial conguration of active users on Rd , denoted by , users located inside C are served by o, users outside this region are served by another base station (or are in outage regime). The power consumed by the battery of the base station o can be divided into two parts: The power dedicated to transmit, receive, decode and encode the signal of any active user. The cumulated power over the whole conguration is then of the form x (x), where is a function to be dened later. The power dedicated to broadcast messages. In order to guarantee that all active users receive these messages, the power must be such that the farthest user in the cell is within the reception range (if the system performs power control) or all the cell is within reception range (if the system does not performs power control). Thus,
78
6. Generality and basic model
BS
MS
Uplink and downlink
Broadcast
Figure 6.1: Power consumption model.
79 the power is a function of maxxC |x| where |x| is the Euclidean norm of x (the power is equal to 0 if C = ). This function is constant if power control is not performed. It follows that the total consumed power is given by: P ( ) = PA ( ) + PB ( ), (6.1)
where PA ( ) = x (x) and PB ( ) = ( ), = maxxC |x| if C = and = 0 if C = (the subscript A stands for "additive" and B stands for "broadcast"). For a very simple propagation model (without fading and shadowing), the Shannons formula states that for a receiver located at x, the transmission rate is given by W log2 (1 + Pe l(x)), where W is the bandwidth, Pe is the transmitted power and l(x) is the pathloss function. Generally, the function l : Rd [0, ] takes the form l(|x|) where l : [0, ] [0, ] is a non decreasing function. This implies that in order to guarantee a minimum rate at position x, Pe must be proportional to l(x). Thus, it is sensible to choose as (x) = a.l(|x|)1{xC } with a > 0. The function is chosen as (x) = b.l(|x|)1{xC } , If power control is performed; b.l(Rinf ), If power control is not performed.
We can divide models for path loss into two categories: singular path loss model l(x) = K |x| where is the exponent path loss parameter and K is a positive constant. non-singular path loss models like l(x) = K (r0 |x|) or l(x) = K (1 + |x| )1 . More generally, we make the following assumption: Assumption 2. The transmitted power depends only on the distance to the base station (x) = (|x|)1{xC } . Furthermore, and are continuous non decreasing function on R+ . We denote (x) = (|x|)1{xC } . This implies that and are always bounded function. Apart from the above model for power consumption, we can dene a power d consumption as a general functional depending on the conguration of users PG : R [0, ]. If = (t , t 0) is a process of time varying congurations, the total consumed energy between time 0 and time T is given by
T
JT := JT (, T ) =
0
P (s )ds
T
As previously, we also dene JA and JB by:

T
JA := JA (, T ) =
0
PA (s )ds and JB := JB (, T ) =
0
PB (s )ds
The same denition for JG (, T ) if the system applies power consumption PG (.). Also we denote C (r ) = C B(o, r ) and C (r ) = C B(o, r ). For a conguration , we denote x the point of such that |x | = (if there are more than one point then xnu is randomly chosen among these points).
80
6.3
Model for mobility of users
In this section, we introduce the mobility models for users. Consider the functional space D ([0, ), Rd ) of Cdlg function on Rd equipped with the Skorohod topology (see, for example [53], page 369). It is well known that D ([0, ), Rd ) is a Polish space. The subset D0 ([0, ), Rd ) = {f D ([0, ), Rd ), f (0) = o} equipped with the Skorohod topology is also a Polish space. We consider a probability distribution PM of a random variable M = (M (t), t [0, )) dened on the associated Borelian eld of the space D0 ([0, ), Rd ). Each realization of M can be represented as a Cdlg trajectory of on Rd . Also, this probability is completely determined by the distributions of nite marginal distributions P (M (t1 ) ., ..., M (tn ) .) (t1 , ..., tn > 0). In some situation, for convenience we can assume that M (t) = o for t < 0. M is said to satisfy the property T if P(M (t1 ) = M (t2 )) = 0 for any 0 t1 < t2 . If mobility is considered, then each user is associated with a mobility process on Rd . We make the following assumption: Assumption 3. Motion trajectories of users are i.i.d mutually independent and have the same distribution as that of M . Motion trajectories of users do not depend on the initial position of users. More precisely, a user i initially located at xi is associated with an independent version of M , namely Mi and an arrival time Ti . This user will move during its sojourn along Mi , i.e the position of this user at time t Ti is x + Mi (t Ti ). The random processes (Mi )iN are mutually independent. Examples for mobility model are as follows: Motionless users: M (t) = o, t R. Brownian motion users: M (t) = c(t)Bd (t) where c(t) R is a continuous function in [0, ) and Bd is a standard Brownian motion on Rd . Completely aimless users: M (t) = tv where the speed of user v is random whose direction is chosen randomly and uniformly over the d-dimensional unit sphere and |v | is a positive random variable. Combination of two above models: M (t) = tv + c(t)Bd (t). High mobility regime: let > 0 be a small parameter, the high mobility regime consists of considering the mobility process (M/)(t) = M (t)/ and we want to study the behavior of the system when . The high mobility regime of the completely aimless mobility model with constant speed |v | is the same as considering |v | , i.e when the users speed is very high.
6.4
Basic model
In this section, we present a very basic model for energy consumption of cellular network. We assume that for each time t, t follows a Poisson point process of intensity dx and the cell C is circular centered at the origin. Furthermore, (x) = (|x|) = a |x| 1{xB(o,R)}
81
R o
Figure 6.2: Basic model.
and (r ) = br 1{xB(o,R)} . We are interested in the average energy E [JT (, T )] that the BS consumes during the period [0, T ). As the conguration at any time follows the same distribution, we have E [JT (, T )] = T E [P (0 )] = T E [PA (0 )] + T E [PB (0 )]. Therefore, it is sucient to nd E [PA (0 )] and E [PB (0 )]. Since PA is linear functional of 0 , thanks to the Campbells theorem and the lemma 28 we can calculate the expectation of PA as follows:
E [PA (0 )] = aVd
R +d +d
Lemma 28. Denote Vd = then
d/2 +1) ( d 2
the volume of a ball of radius 1 in Rd , and Vd = dVd
B(o,R)
|x|k dx = Vd
R k +d k+d
for all real k > d.
82 Proof. We have: |x|k dx = =

0 R 0 R
r k d Vd r k Vd r k+d1 dr
B(o,R)
= Vd
R k +d k+d
We are now interested in PB . Lemma 29. Let be a Radon measure on Rd absolutely continuous with respect to the Lebesgue measure, let be a Poisson point process of intensity on Rd then the CDF of is given as: F and its PDF is given by p

(r ) = e (C (r)) ,
(r ) =
d (C B(o, r )) (C (r)) e dr
In particular, if d (x) = dx and C = B(o, R) then: F and p

(r ) =
eVd (R 1,
d r d )
, 0 r < R; R r.
(r ) =
Vd r d1 eVd (R 0,
d r d )
, 0 r < R; R r.
Proof. We have, as is a Poisson point process: F
(r ) = P C (r ) = = e (C (r))
for r R. The PDF of then follows. Now if d (x) = dx and C = B(o, R) then (C B(o, r )) = Vd (Rd r d ), then the expressions for CDF and PDF of follows. Remark 1. If the cell C has the regular npolygon (n = 3, 4 or 6 in a regular network) on R2 (d = 2) with circumradius R, and d (x) = dx then the CDF of in the previous lemma can be expressed as follows: exp (R2 r 2 ) , if 0 r < R cos n; R cos n R cos 2 exp R n arccos r + n sin 2 arccos r n r2 , 2 F (r ) = r < R; if R cos n 1, if R < r .
83 Following the above lemma, the moments of PB are given as mn [PB (0 )] = bn

0 R
r n Vd r d1 eVd (R
d r d )
dr
For simplicity, let

u
Hd (u, v, ) = dv
0
r r d1 ev(u
d r d )
dr
(6.2)
where 0 then, mn [PB (0 )] = bn Hd (R, Vd , n ). In particular, E [PB (0 )] = bHd (R, Vd , ) Consequently, we obtain E [JT (, T )] = T aVd
R +d + T bHd (R, Vd , ). +d
+d +d
R We note that 0 Hd (u, v, ) R so T aVd R +d E [JT ] T aVd +d + T bR . If then E [JB ] T bRd , which is intuitive: as the intensity goes larger, the farthest user is located closer to the border of the cell. Nevertheless, bT R is the consumed energy if the operator wants to broadcast message to all points of the cell, not just all active users in the cell. Thus, it is fair to assume that E [JB (, T )] T b R with b 0. Consider an operator aiming to design the optimal cell radius R to cover a region of total area (volume) S Rd . The number of base station is then V S d . The average total cost dR of the network is the sum of the operation cost during the life time of the network(says T ) and the cost of facilities (base stations). If the cell radius is R then the cost for installation c1 1 of base stations is linear function of R d , say Rd with c1 > 0. The operation cost is assumed to be proportional to the consumed energy. According to results proved above, the operation cost of a base station is of the form a T Rd+ + b T R , which is a increasing function of R (a , b 0). Thus, small cells reduce the total energy consumption in a cellular network. The average total cost for the network is then
Cost(R) =
c1 S a T Rd+ + b T R + d d Vd R R c1 d = a1 T R + b1 T R + d R
It is interesting for operator to minimize the cost function. Assume that one of two parameters a1 and b1 is strictly positive and > d. As Cost(R) > 0 for all R > 0 and limR0 Cost(R) = limR Cost(R) = there exists a minimum for Cost. By dierentiation, the optimal cell radius must be solution of the following equation:
+d a1 T Ropt + b1 T ( d)Ropt = dc1
As the RHS is increasing in T , the optimal cell radius must be a decreasing function of R. This reveals a characteristic of the optimal choice of cell radius. In the economical point of view, to operate a network with longer life time it is preferable to exploit smaller cells system.
84
If b1 = 0, i.e the broadcast part of transmitted power is small comparing to the additive part or the operator does not broadcast then the problem reduces to minimizing a1 T R + c1 . Simple manipulations yields: Rd Ropt = dc1 a1 T
1 +d
That is to say theoretically the optimal cell radius is proportional to (T ) +d . Similarly the network operates only in broadcast, the optimal cell radius will be Ropt = dc1 ( d)b1 T
1
6.5
Conclusion
In this chapter we have proposed a general energy consumption on cellular networks and a model for mobility for users. We have also considered a basic model for energy consumption and in order to minimize the cost of the network we derive some theoretical results on the optimal choice of cell radius.
85
Chapter 7
ON-OFF model
Contents
7.1 7.2 7.3 7.4 7.5 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85 Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86 Motionless case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88 Impact of mobility . . . . . . . . . . . . . . . . . . . . . . . . . . . 93 Special cases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98 7.5.1 7.5.2 7.6 Completely aimless mobility model . . . . . . . . . . . . . . . . . Always on users . . . . . . . . . . . . . . . . . . . . . . . . . . . 98 99
Summary and Conclusion . . . . . . . . . . . . . . . . . . . . . . 102
7.1
Introduction
The previous chapter introduced a new energy consumption model for cellular network. In this chapter, we apply it in a specic scenario. At time t = 0, users are dispatched in the plane according to a Poisson point process. With each of them, is associated a random process with two states ON-OFF (active or inactive) which represents their activity. Moreover, users randomly move. Only "on" users are served by the network. This is the very rst approach we can think of to model the activity of entities in a network. As usual, this assumption reects the fact that a user can disconnect from a wireless network to switch to a new wire network or the channel from the BS to the user can become so bad so that the connection is no longer possible,... . Although this approach seems to be simple at the rst sight, it is still a complex object to study and need a lot of calculation. The ON-OFF model is well studied in the queueing literature. It is used to explain phenomenons called heavy tails, self-similarity and long range dependence of trac observed by measurement in many types of networks, see for example [54] and references therein. We do not address this problem here in this thesis. This chapter is organized in the following way. The section 7.2 describes the model. Section 7.3 presents analysis in the case where users are motionless. Section 7.4 investigates the impact of mobility. Section 7.5 considers some special cases of mobility model, including the completely aimless mobility model.
86
7. ON-OFF model
7.2
Model
An ON-OFF process on the real line alternates between values 1 (for on state) and 0 (for OFF state). ON-periods (and OFF-periods) are i.i.d positive random variables. Furthermore, the sequences of ON-periods and OFF-periods are independent. Each realization of an ON-OFF source is a Cdlg function. An ON-OFF process is called exponential if ON-periods and OFF-periods are exponential distributed. More precisely, consider an ON-OFF sources (I (t), t R) such that the ON-periods 1 are continuous positive random variable of mean 1 and the OFF-periods are continuous 1 positive random variable of mean 0 and denote by U and V the generic ON-period and OFF-period. We can write: I (t) =
i=
1{T2i t<T2i+1 } .
where ... < T3 < T2 < T1 < T0 < T1 < T2 < T3 < ... such that (T2i T2i1 ) i= are i.i.d and have the same distribution as V , and (T2i+1 T2i ) are i.i.d and have the same i= distribution as U . We can assume that T0 T1 0. I can be seen as a random variable on D (R, R) with probability measure dPI . We assume that I is stationary. From [55] 1 0 1 and P(I (t) = 1) = 0 0 for all t. for example, we have P(I (t) = 1) = 0 +1 +1 Furthermore, we have P (I (t) = 1t [0, u)) = 1 E [U ]
u
P (U > s) ds.
for all u > 0. With a little abuse of notation, we write j1 ,...,jn (t1 , ..., tn ) = P (I (ti ) = ji i = 1..n) T where j1 , ..., jn {0, 1} and t1 , ..., tn R. Let A(T ) = 0 I (t) dt. Since the process I is stationary, we have E [A(T )] = 1 T . In the case of exponential ON-OFF process, the expressions for j1 ,...,jn (t1 , ..., tn ) and the moments of A(T ) are given in the appendix. Lemma 30. We have, for all T
n n 1 T mn [A(T )] T n .
Proof. Since A(T ) T a.s. we have mn [A(T )] T n . Now mn [A(T )] (E [A(T )])n = n T n by CauchySchwarz inequality. 1 Let d(x) be a nite Radon measure on Rd , absolutely continuous with respect to the Lebesgue measure. We make the following assumptions: Assumption 4. The positions of users at t = 0 follow a Poisson point process 0 = {Xi }i1 of intensity measure d(x). User i is associated with an ON-OFF process of activity (Ii (t), t R),i.e users are active during their ON-periods and are inactive during their OFF-periods. The activity processes of users are assumed to be i.i.d and have the same distribution as that of (I (t), t 0). Following the above assumptions, the conguration of active users at time t is
M t = i 1
1{Ii (t)=0} Xi +Mi (t)
87
Time
0 1 2 3 4 5 6 User
Figure 7.1: Illustration of the model, each user is associated with a ON-OFF process and a mobility process.
88
7. ON-OFF model
The system can be described by a Poisson point process on Rd D (R, R) D ([0, +), Rd ) I,M = {(Xi , Ii , Mi )}i1 of intensity d(x) dPI dPM . The consumed energy is dened in the same way as in the previous chapter for the time varying conguration process M = ( M (t), t 0). In particular, the additive part of consumed energy can be rewritten as: JA ( M , T ) =
i 1 0 T
Ii (t)(Xi + Mi (t)) dt
and the broadcast part is: JB ( M , T ) =

0 T M ( t ) dt
The total consumed energy is JT ( M , T ) = JA ( M , T ) + JB ( M , T ). When users are motionless, i.e Mi (t) = o, the system is described as a Poisson point process I = {(Xi , Ii )}i1 of intensity measure d(x) dPI . In this case, we drop the superscript. Thus, the conguration of users at time t is:
n
t =
i=1
1{Ii (t)=0} Xi
The additive part of consumed energy is

T
JA (, T ) =
i 1
(Xi )
0
Ii (t) dt
and the broadcast part is:

T
JB (, T ) =
0
( t ) dt
The total consumed energy in this case is JT (, T ) = JA (, T ) + JB (, T ). In the next two sections we present analytical results on the motionless case and the general case.
7.3
Motionless case
In this section, we assume that users are motionless. We derive analytical expressions for the moments of JA (, T ), JB (, T ), JT (, T ) in this case. Theorem 31. The moments of JA (, T ) are given by: mn [JA (, T )] = Bn m1 [A(T )] (x) d(x), ..., mn [A(T )] n (x) d(x) , (7.1)
Rd
Rd
89 and the central moments of JA (, T ) are given by: cn [JA (, T )] = Bn 0, m2 [A(T )] 2 (x) d(x), ..., mn [A(T )] n (x) d(x) (7.2)
Rd
Rd
In particular, the expectation of JA (, T ) is given as: E [JA (, T )] = 1 (x) d(x).

Rd
(7.3)
Proof. As JA (, T ) is a linear functional of I which is a Poisson point process, we then apply Theorem 7.4 and note that:
T n
f (x)
Rd D (R,R)) 0
I (t) dt
d(x) dPI =
f n (x) d(x)E
Rd 0
I (t) dt
= mn [A(T )]
Rd
f n (x) dx
We see that, from the above theorem, the expectation of JA (, T ) depends on the distribution of ON-periods and OFF-periods only by the activity rate 0 /1 . Corollary 4. If C = B(o, R), (x) = a |x| 1{xB(o,R)} and d(x) = dx then: mn [JA (, T )] = Bn aVd m1 [A(T )] R +d Vd a2 m2 [A(T )] R2 +d V an mn [A(T )] Rn +d , , ..., d +d 2 + d n + d

cn [JA (, T )] = Bn In particular,
0,
Vd a2 mn [A(T )] R2 +d V an mn [A(T )] Rn +d , ..., d 2 + d n + d
E [JA (, T )] = Proof. We have that
1 daR +d . ( + d)( d 2 + 1)
k +n
B(o,R)
|x|k dx = Vd R k +n with k > n. The proof is, thus, completed.
Applying Theorem 10, an error bound for Gaussian approximation of JA (, T ) is found as follows: Theorem 32. Let JA (, T ) =
JA (,T )E[JA (,T )] V[JA (,T )]
then for any u we have:

Rd Rd
P JA (, T ) > u Q(u)
m3 [A(T )] m2 [A(T )]
3 (x) d(x)
3 2
2 (x) d(x)
90
7. ON-OFF model
2 T 2 and mm3A(T ) T 3 , so the bound As shown in Lemma 30, T 2 m2 [A(T )] 1 decays as (1) as T . We obtain a less sharp bound but depending only on the activity rate 1 but not on the distribution of ON-periods and OFF-periods and on T : Rd 3 1
P JA (, T ) > u Q(u)
3 (x) d(x)
3 2
2 Rd (x) d(x)
As already noted in the case of exponential ON-OFF source, when T goes to innity, 2 T 2 and m [A(T )] 3 T 3 . Consequently, in this case the bound has the m2 [A(T )] 1 3 1 following limit when T :
Rd
3 (x) d(x) 2 (x) d(x)

3 2
Rd
Now let (x) = dx, the bound becomes: m3 [A(T )] m2 [A(T )]

Rd
3 (x) dx 2 (x) dx
3 2
Rd
1 which decays as ( ) as .
Theorem 33. The joint distribution of ( t1 , ..., tn ) is given by:

n
F(
t1 ,..., tn
) (u1 , ..., un ) =
i=1
(1)i1
1k1 <...<kin
e1,1,...,1 (tk1 ,...,tki )(C (max{uk1 ,...,uki }))
In particular, F
t
(u) = e1 (C (u))
(7.4)
The nth order moment of JB (, T ) are given by: mn [JB (, T )] = dt1 ... dtn (u1 )... (un ) dF(
t1 ,..., tn
[0,T )n
(R+ )n
) (u1 , ..., un )
Proof. Let Ai be the event {|ti | ui }. Simple joint probability rule gives us
n n
P
i=1
Ai
=
i=1
(1)i1
1k1 <...<ki n
P i j =1 Aki
i Thus, we must evaluate P i j =1 Aki . The event j =1 Aki is equivalent to the event: there is no user on C being active at all instances tk1 , ..., tki having distance to the origin larger than max{uk1 , ..., uki }. We recall that the ON-OFF processes of users are i.i.d, so the users being active at all these instances tk1 , ..., tki form a Poisson point process of intensity 1,...,1 (tk1 , ..., tki ) d(x) and thus 1,1,...,1 (tk1 ,...,tki )(C (max{uk1 ,...,uki })) . P i j =1 Aki = e
91 The expression for the joint distribution of ( t1 , ..., tn ) follows. Now we can write:
T n
mn [JB (, T )] = E
0 T
( t ) dt
T
=
0
...
0
E (|t1 |)... (|tn |) dt1 ... dtn

(R+ )n
=
[0,T )n
dt1 ... dtn
(u1 )... (un ) dF(
t1 ,..., tn
) (u1 , ..., un ).
This completes the proof. Theorem 34. The second order moment of JT (, T ) is given by:
2
m2 [JT (, T )] =
m1 [A(T )] +
[0,T )2
(x) d(x)
Rd
+ m2 [A(T )]
2 (x) d(x)
Rd
dt1 dt2
(R+ )2
(u1 ) (u2 ) dF(
t 1 , t 2
) (u1 , u2 ) (7.5)
+2 RHS of (7.6). Proof. We have: E (JT (, T ))2 = E (JA (, T ) + JB (, T ))2
= E (JA (, T ))2 + E (JB (, T ))2 + 2E [JA (, T )JB (, T )] We have already expressions for E (JA (, T ))2 and E (JB (, T ))2 , thus it remains to evaluate E [JA (, T )JB (, T )].
T T
E [JA (, T )JB (, T )] = E
0 T
PA (t ) dt
0 T 0
PB (t ) dt
=
0
E PA (t ) ( s ) dt ds
We have to evaluate E [PA (t ) |s |]. We can do it by determining the conditional expectation E [PA (t )| s = r ] then deconditioning on the distribution of |s |. Now consider the event "|s | = r ". Given this event, the conguration of users being active at t is a Poisson point process of intensity measure 10 (t, s)1{xC (r)} d(x) + 1 1{xC (r)} d(x) plus the user at distance r if it is active. Consequently, using Campbells formula we have: E [PA (t )PB (s )| s = r ] = 10 (t, s) (r ) +1 (r )
C (r )
(x) d(x)
C (r )
(x) d(x) + p11 (t, s)(r ) (r )
Note that the distribution of s is the same as that of 0 , we nally obtain an expres-
92 sion for E [JA (, T )JB (, T )]:

T T
7. ON-OFF model
E [JA (, T )JB (, T )] =
0 0 T
10 (t, s)
0 T 0 T 0 0 0
(r )
C (r )
(x) d(x) dF (x) d(x) dF

0
(r ) ds dt
+ 1
0 T
(r )
C (r )
(r ) ds dt (7.6)
+
0
11 (t, s)(r ) (r ) dF
(r ) ds dt
We have obtained closed form formulas for V [JB (, T )] and V [JT (, T )], however it requires to solve triple integrals. We now present bounds for V [JB (, T )] and V [JT (, T )], which requires only one easily computed integral. Theorem 35. We have: V [JB (, T )] m2 [A(T )] and m2 [A(T )]
Rd
2 (x) d(x),
Rd
(7.7)
2 (x) d(x) V [JT (, T )] m2 [A(T )]
((x) + (x))2 d(x). (7.8)

Rd
Proof. As (7.7) is covered by (7.8) by setting (x) = 0, it is sucient to prove (7.8). Since JT (, T ) is a linear functional of the Poisson point process I we can apply the corollary 1. By denition, adding a point (X, I ) to will increase the consumed power for at most (X ) + (X ) at each instance where the new user X is active. Consequently, T 0 D(X,I ) JA (, T ) ((x) + (x)) 0 I (t) dt. It follows that: V [JT (, T )] ( (x) + (x)) d(x)E
Rd 0 2 T 2
I (t)
where the last expectation is taken with respect to the distribution of I . Now we apply Theorem 5. We consider two functionals JA (, T ) and JB (, T ). Adding a point (X, I ) will not decrease both the functionals, so: m2 [A(T )]
Rd
2 (x) d(x) = V [JA (, T )] V [JA (, T ) + JB (, T )] = V [JT (, T )] .
From theorem 5, we also note that V [JT (, T )] V [JA (, T )] + V [JB (, T )] . Corollary 5. If C = B(o, R), (x) = a |x| 1{xB(o,R)} , (x) = b |x| 1{xB(o,R)} (a, b > 0) and d(x) = dx then: V [JB (, T )] V [JT (, T )] a2 m2 [A(T )] Vd Rd+2 d + 2 (a + b)2 m2 [A(T )] Vd Rd+2 d + 2
93 The following theorem gives an upper bound on the distribution of JT (, T ): Theorem 36. Assume that (x) + (x) K for all x Rd , let 2 = m2 [A(T )] then P (JT (, T ) > E [JT (, T )] + u) exp for all u > 0. Proof. As proved above, 0 DX,I JT (, T ) ((x) + (x)) 0 I (t) dt T K . Thus, applying Theorem 7 (or its corollary 1) we obtain the desired result. By setting (x) = 0 or (x) = 0 in the above theorem, we can derive upper bound on the distribution of JA (, T ) and JB (, T ): Corollary 6. Let 2 A = m2 [A(T )] 2 B = m2 [A(T )] ((x))2 d(x)
Rd T
( (x) + (x))2 d(x)

Rd
T 2K 2 g 2
uT K 2
( (x))2 d(x).
Rd
Assume that (x) K for all x Rd then for any u > 0, P (JA (, T ) > E [JA (, T )] + u) exp T 2K 2 g 2 A uT K 2 A .
Similarly, assume that (x) K for all x Rd then for any u > 0, P (JB (, T ) > E [JB (, T )] + u) exp T 2K 2 g 2 B uT K 2 B
7.4
Impact of mobility
To consider the eect of mobility, we always assume that d(x) = dx.

M is a Poisson point process of intensity measure dx for all t. Lemma 37. t 1
Proof. Consider the point process i1 Xi +Mi (t) . By the displacement theorem, it is a Poisson point process of intensity measure dt (x) characterized by: t (A) = =
Rd Rd
P(x + M (t) A) dx dx
Rd
pM (t) (y )1{x+yA} dy
Rd
=
Rd
pM (t) (y ) dy
1{x+yA} dx
= ld (A).
M = Thus, it is a Poisson point process of intensity dx. Now by thinning property, t i1 1{Ii (t)=1} Xi +Mi (t) is a Poisson point process of intensity 1 dx.
94
7. ON-OFF model
Theorem 38. For any power allocation policy PG , and for any mobility model M , the expectation of energy consumed is the same as in motionless case, i.e: E JG ( M , T ) = E [JG (, T )] In particular E JA ( M , T ) = E [JA (, T )], E JB ( M , T ) = E [JB (, T )] and E JT ( M , T ) = E [JT (, T )].
M and follow the same distribution (Poisson point processof Proof. As for each t, t t T M ) dt and E [J (, T )] = intensity 1 dx). Consequently, E JG ( M , T ) = 0 E PG (t G T E [ P ( )] d t must be equal. G t 0
For a non negative function f Ln (Rd ) and t1 , ..., tn R and furthermore we assume that f (x) = 0 if x Rd /C . Dene M n (f, t1 , ..., tn ) = Lemma 39. We have M n (f, t) =
Rd
E [f (x + M (t1 ))...f (x + M (tn ))] dx. f (x) dx and: f n (x) dx

M/
Rd
M n (f, t1 , ..., tn )
Rd
Moreover, if M has the property T and n 2 then n (f, t1 , ..., tn ) 0 as 0 with M/ n 2. If f (x) = a1{xC } with a > 0 and n 2 then n (f, t1 , ..., tn ) is decreasing function of . Proof. Note that for all y Rd , Rd f (x + y ) dx = theorem and CauchySchwarz inequality: M n (f, t1 , ..., tn ) = E E =
Rd Rd Rd
f (x) dx. We have, by Fubinis
f (x + M (t1 ))...f (x + M (tn )) dx ((f (x + M (t1 )))n dx) n ... ((f (x + M (tn )) dx)n ) n dx
1 1
Rd
f n (x) dx,
and it is easy seen that inequality occurs if n = 1. M (t2 )| Since for each realization of M one has |M (t1 ) as M (t1 ) = M (t2 ) (a.s), M (t2 ) M (t1 ) thus for small enough, one of 2 points x + and x + must be located outside C, M (tn ) M (t1 ) M (tn ) t1 ) then f (x + M ( )...f (x + ) = 0. It means that f (x + )...f (x + ) 0 (a.s) n and it is bounded by (sup f ) , so its expectation tends to 0. By dominated convergence theorem, we obtain that E
Rd
x+
M (t1 )
...f
x+
M (tn )
dx 0.
Now assume f (x) = a1{xC } and n 2, we have: M/ n (f, t1 , ..., tn ) = E = E

C Rd
1{x+ M (t1 ) ,...,x+ M (tn) C } dx

1{x+ M (t2 )M (t1 ) ,...,x+ M (tn)M (t1 ) C } dx .
95 Consider 0 < 1 < 2 , due to the convexity of C , for each 1 < i n if x +

M (t1 ) x+ M (ti ) 2 M/
M (ti )M (t1 ) 1
then
C . Since C is bounded, it implies that 1{x+ M (t2 )M (t1 ) ,...,x+ M (t2)M (t1 ) C } (f, t1 , ..., tn ) is also decreasing function of
n
is decreasing function of . It means that n . Now let M n (f, T ) = Lemma 40. We have:
Rd
T 0
f (x + M (t))I (t) dt
dx.
M n (f, T ) mn [A(T )] If M has the property T then f a > 0 then

M/ f (T, n) M/
Rd
f n (x) dx
is decreasing function of for n 2.

n
(T, n) 0 as 0 for n 2. If f (x) = a1{xC } with
Proof. We can write, by the same way as previously done and using Lemma 39: M n (f, T ) = =
Rd T 0 T T T
...
Rd 0 T 0 T
E
i=1 n
(f (x + M (ti ))I (ti ))

n
dt1 ... dtn dx I (ti ) dt1 ... dtn dx dx dt1 ... dtn
...
0
E
i=1 n
f (x + M (ti )) E
i=1 n
=
0 T
...
0 T
E
i=1 n
I (ti )
E
Rd i=1
f (x + M (ti ))
=
0 T
...
0 T
E
i=1 n
I (ti ) M n (f, t1 , ..., tn ) dt1 ... dtn I (ti )

i=1 T T
...
0 n 0
Rd
f n (x) dx dt1 ... dtn

n
f (x) dx
Rd 0
...
0 T
E
i=1
I (ti )
n
dt1 ... dtn
=
Rd
f (x) dxE
0
I (t) dt
= mn [A(T )]
Rd
f n (x) dx
The rst part of lemma is, thus, proved. Now the second part follows the fact that M/ M/ n (f, t1 , ..., tn ) 0 as 0; n (f, t1 , ..., tn ) is bounded by Rd f n (x) dx and the dominated convergence theorem. If f (x) = a1{xC } and n 2 then f decreasing function of because
M/ n (f, t1 , ..., tn ) M/
(T, n) is a
is.
Theorem 41. The moments of JA ( M , T ) are given by mn JA ( M , T ) cn JA ( M , T )

M M = Bn (M (T, 1), (T, 2), ..., (T, n)) M = Bn (0, M (T, 2), ..., (T, n))
96 Mobility reduces moments of JA ; i.e
7. ON-OFF model
mn JA ( M , T ) mn [JA (, T )] , and cn JA ( M , T ) cn [JA (, T )] Furthermore, we have: E exp JA ( M , T ) E [exp {JA (, T )}] (7.9)
for all R. If M has the property T then the central moments of JA goes to 0 in high mobility regime; i.e
n
cn JA ( M/ , T ) 0, and mn JA ( M/ , T )
(x) dx
Rd
as 0
If (x) = a1{xC } with a > 0, M has the property T and n 2 then mn JA ( M/ , T ) , cn JA ( M/ , T ) are decreasing functions of . Proof. The proof is similar to that of Theorem ??. As JA ( M , T ) is a linear functional of I,M we can apply Theorem . We derive that
T n
f (x + M (t))I (t) dt
Rd D (R,R)D ([0,+),Rd ) T 0 n
dx dPI dPM
= =
Rd
E
0
f (x + M (t))I (t) dt
dx
M (T, n)
Hence the expressions of mn JA ( M , T ) follow and cn JA ( M , T ) . Now using results of Lemma 40 and the fact that a Bell polynomial has non negative coecients, we have mn JA ( M , T ) mn [JA (, T )] , and cn JA ( M , T ) cn [JA (, T )]. Now assume M has the property T and following lemma 40, we have, as 0 and n 2: cn JA ( M/ , T ) mn JA ( M/ , T ) Bn (0, 0, ..., 0) = 0 Bn (1 (x) dx, 0, ..., 0) =
Rd n
(x) dx
Rd
Now, following the Laplace functional of Poisson point process and the Jensens inequality, we have: E exp JA ( M , T ) = exp
Rd
E e
T 0
(x+M (t))I (t) dt

T 0
dx 1 dx dt 1 dx dx dx
= exp
Rd
EI EM e EI EM
0 T
(x+M (t))I (t) dt
exp
Rd
I (t) (x+M (t))A(T ) e A(T )

Rd
= exp EI EM
0 T
I (t) A(T ) I (t) A(T )
e(x+M (t))A(T ) 1 e(x)A(T ) 1 dx
= exp EI EM
0
dx dt
Rd
= exp EI
Rd
e(x)A(T ) 1
= E [exp {JA (, T )}]
97 If (x) = a1{xC } with a > 0 and n 2 then mn JA ( M/ , T ) , cn JA ( M/ , T ) are

M/
decreasing functions of because f
(T, n) is.
Theorem 42. The variance of JB ( M , T ) and JT ( M , T ) are bounded as follows: V JB ( M , T )

M M (T, 2) V JT ( , T )
M + (T, 2)
M (T, 2)
Moreover, if M has the property T then in high mobility regime, the variance of JB and JT tends to 0, i.e V JB ( M/ , T ) , V JT ( M/ , T ) 0 as 0 Proof. The proof is similar to that of Theorem 35. We need only to prove results for JT ( M , T ), the results for JB ( M , T ) follows by setting (x) = 0. We apply Corollary 1. By denition, 0 D(X,I,M ) JA ( M , T ) Thus, V JT ( M , T ) M + (T, 2)
T 0
((X + M (t)) + (X + M (t))) I (t) dt
We also note that adding a point (X, I, M ) to M will not decrease JA ( M , T ) and JB ( M , T ), so: V JT ( M , T ) V JA ( M , T ) = M (T, 2)
M/
Finally, the convergence to 0 of V JT ( M/ , T ) follows the fact that + (T, 2) tends to 0 as tends to 0. The above results say that, when users move the total consumed energy by a base station does not change in average, and the moments and central moments of the additive part are reduced. Moreover, when users move very fast, the consumed energy during a time period is almost constant. We can see this fact as a consequence of weak central limit d theorem. When users move faster, the conguration of users takes more "value" on R during a same period of time, thus converge faster to the mean. We nd an error bound for Gaussian approximation of JA ( M , T ) as follows: Theorem 43. Let JA ( M , T ) =
M JA ( M ,T )E[JA ( M ,T )] V[JA ( M ,T )]
then:
P JA ( , T ) > u Q(u) Proof. This result is consequence of Theorem 2.17.
M (T, 2) M (T, 3)
3/2
Theorem 44. Assume that (x) + (x) K for all x Rd , then P JT ( M , T ) > E JT ( M , T ) + u exp for all u > 0. T 2K 2 g M + (T, 2) uT K M + (T, 2)
98
7. ON-OFF model
Proof. The proof is similar to the proof of Theorem 36, and it is the consequence of Corollary 2. By setting (x) = 0 or (x) = 0 in the above theorem, we can derive upper bounds on the distribution of JA ( M , T ) and JB ( M , T ): Corollary 7. Assume that (x) K for all x Rd , then for any u > 0, P JA ( M , T ) > E JA ( M , T ) + u exp T 2K 2 g M (T, 2) uT K M (T, 2) .
Assume that (x) K for all x Rd , then for any u > 0, P JB ( M , T ) > E JB ( M , T ) + u exp T 2K 2 g M (T, 2) uT K M (T, 2)
7.5
Special cases
In this section we consider some special mobility model M .
7.5.1
Completely aimless mobility model
Consider the completely aimless mobility model M with constant speed, i.e M (t) = tv where |v | = constant and the direction of v is uniformly distributed. Lemma 45. Let f (x) be a positive measurable function on Rd such that f (x) = 0 for 1 x Rd /C , f (x) c1 for all x C and f (x) c2 for all x C/B (o, R 4 ) where c1 , c2 > 0 are constant then M n (f, T ) = and M n (f, T ) = (T ), T for n 2. Proof. (see Figure 7.2) Assume that T that if t M n (f, T )
R1 4|v|
R 1 and x C then x + tv C/B (o, R 4 ). As a consequence,: 1 T 4|v | 1 t1 + 4|v | R 1 t1 + 4|v | R
1 |v |n1
, |v |
2R |v | .
R1 1 Let C = B (0, 3R 4 )/B (0, 2 ) and we note
cn 2E
dx
0
R
dt1
R
t1
dt2
R
t1
I (t1 ) I (tn ) dtn
cn 2 ld (C )
1 T 4|v |
dt1
1 t1 + 4|v |
dt2 ...
R1 4|v |
1 t1 + 4|v |
t1
R1 4|v |
t1
P (I (t1 ) = = I (tn ) = 1)) dtn R1 ) 4 |v | dtn
cn 2 ld (C )
R1 T 4 |v | R1 4 |v |
dt2 ...
0 n 1
P I (t) = 1t [0,
R1 4|v |
= cn 2 ld (C ) T
R1 4 |v |
1 E [U ]
P (U > t) dt.
99 Similarly, note that for |t| must be outside the cell, thus M n (f, T ) cn 1E
T 2R |v|
then for all x, at least one of two points x or x + vt

R t1 + 2 T |v | R t1 2 |v | + R t1 + 2 T |v | R t1 2 |v | +
dx
C 0
dt1 2R |v |
dt2
I (t1 )...I (tn ) dtn
cn 1 ld (C ) T Remarking that
1 E[V ]
R1 4|v |
4R |v |
n 1
P (V > t) dt 1 as |v | , we obtain the desired result.
Theorem 46. Consider the completely aimless mobility model M with constant speed |v | then
1 1. V JA ( M , T ) and V JT ( M , T ) decay as ( |v | ) as |v | and decay as O (T ) as M M V[J ( ,T )] V[J ( ,T )] and E[J T decay as ( 1 ) as |v | T . The ratios E[J A ( M ,T )] ( M ,T )]
A T
1 and ( ) as T . T
|v |
2. The error bound of Gaussian approximation in Theorem 43 decays as (|v |2 ) as 1 ) as T . |v | and decays as ( T Proof. From the assumption 2 we have (x) c1 for all x C , (x) c2 for all x 1 C/B (o, R 4 ) and 0 (x) c3 for all x C for some nite positive constants c1 , c2 , c3 . Now, according to Theorem 42:
M M M M (T, n) = V JA ( , T ) V JT ( , T ) + (T, n).
The results then follow from Lemma 45. We also see that, if |v | is small, the variance of JT ( M , T ) is proportional to |T v| while 2 in the motionless case, it is (T ). We also notice that mobility makes the Gaussian 1 approximation of JA more accurate when T is large as the bound decays as ( ) instead T of (1) in the motionless case. In the motionless case the position of users are always xed over time, only their state change, thus the conguration of active users can take d only some possible values on R . So it is intuitive that one cannot guarantee that the Gaussian approximation is good if T is large. On the contrary, in the mobility case, the d conguration of active users can take all possible values on R . When T grows larger, it take more values. Thus, Gaussian approximation is better when T grows larger. Quite surprisingly, when |v | is large, the variance of JA tends to 0 but the bound on Gaussian approximation does not decrease. Remark 2. From the above proof and using the properties of Bell polynomial, we can prove that c3 JA ( M , T ) = ( |v1|2 ), c4 JA ( M , T ) = ( |v1|2 ), c5 JA ( M , T ) = ( |v1|3 ),...
7.5.2
Always on users
We now consider the following special case: users are always ON, i.e I (t) = 1 for all t. We always assume the mobility of users. The conguration of users at time t is
M t = Xi +Mi (t)
100 x + vt
7. ON-OFF model
C x + vt
x o R1 x
Figure 7.2: Illustration of the proof of Theorem 46.
In this case, the analysis is simpler than in the ON-OFF case and the results obtained in the previous section can be inherited. Let
M Hn (f, u) = u n
E
Rd 0
f (x + M (t)) dt
dx
(7.10)
M (f, u) is a version of M (f, T ) for I (t) = 1, t R. In particular, the We see that Hn n M moments of JA ( , T ) is given as
mn JA ( M , T ) cn JA ( M , T )
M M = Bn (0, H2 (, T ), ..., Hn (, T )) cn [JA (, T )] .
M M M = Bn (H1 (, T ), H2 (, T ), ..., Hn (, T )) mn [JA (, T )]
M (f, T ) in the case where In the following lemma, we provide explicit expression for Hn d = 1 or 2, f (x) = a1{|x|R} :
Lemma 47.
1. If d = 1, C = [R, R] and f (x) = a1{|x|R} with a > 0 then:

M Hn (f, u)
2Run an
n1 n+1 | v | an , n+1 u u(2R)n n n1 (2R)n+1 n a n +1 |v|n a , |v|n1
2R |v | ; 2R |v | .
(7.11)
101
2 R2 r 2
r o R
Figure 7.3: Illustration of the proof of Lemma 47.
2. If d = 2, C = B(o, R) and f (x) = a1{|x|R} with a > 0 then:

M Hn (f, u) =
2u(2R)n Un (r0 ) n 2(n 1) (2R)n+1 Un+1 (r0 ) n a a n+1 |v |n |v |n1 2(n 1) n+1 u R(1 r0 ) |v | an +4un R2 (U1 (1) U1 (r0 ))an n+1 2u(2R)n Un (1) n 2(n 1) (2R)n+1 Un+1 (1) n a a n+1 |v |n |v |n1
1 0
if u
2R |v |
and
M Hn (f, u) =
if u
2R |v |
where Un =
0
1 r2
u
n 2
dr and r0 =
n
u2 |v|2 . (2R)2
Proof. 1) We can assume that a = 1. We have, H M (f, u) = = = =

[0,u)n
E
0 u
1{|x+vt|R} dt
n
dx dx
1{|x+|v|t|R} dt
[0,u)n
dx
1{|x+|v|t1 |,...,|x+|v|tn |R} dt1 ... dtn

dt1 ... dtn
1{|x+|v|t1 |,...,|x+|v|tn |R} dx
=
[0,u)n u
(2R |v | (max{t1 , t2 , ..., tn } min{t1 , t2 , ..., tn }))+ dt1 ... dtn

u u
= n!
0
dt1
t1 u
dt2 ...
tn1 u
(2R |v | (tn t1 ))+ dtn
= n(n 1)
dt1
0 t1
(2R |v | (tn t1 ))+ (tn t1 )n2 dtn .
102
7. ON-OFF model
By some elementary manipulations, the last integral can be easily computed and it is equal to the LHS of equation (7.11). 2) We will use 1) to prove 2) (see gure 7.3). We also assume that a = 1. Firstly we note that the direction of v does not make any impact, so without loss of generality we assume that v follows the direction of the vector (1, 0). We then have: H M (f, u) =
C
1{x+vt2 ,...,x+vtnC } dx
R
= 2
0 R
dr
R2 r 2
R2 r 2
1{|r1 +|v|t2 |,...,|r1 +|v|tn |R2 r2 } dr1
= 2
0
gn (r ) dr.
2R |v |
Here gn (r ) can be calculated by applying 1). If u gn (r ) = 2 R2 r 2 un If u

2R |v |
then we have
n 1 n+1 u |v | . n+1
then gn (r ) =
u(2 R2 r 2 )n n1 (2 R2 r 2 )n+1 , n1 n+1 |v |n |v | n1 n+1 n 2 2 |v | , 2 R r u n+1 u
0 r r0 ;
R r r0 .
This concludes the proof. From the above lemma, we see that in the case where (x) = constant, the low (and R 2R high) mobility regime can be characterized as T 2 |v| (and T |v| ). For a numerical example we choose the simplest case where we have exact expressions: d = 1, = 0.02 (users/m) 2R = 100(m), (x) = 1 and M (t) = vt with v = |v | or v = |v | with equal probability and we consider only JA . In the gure 7.4 we plot the
V[J ( M ,T )]
in function of |v | in the case T = 106 (x) 11(days) and T = 107 (s) ratio E[J A M A ( ,T )] 115(days). As expected it is a decreasing function of |v |. We see that in the low mobility regime, the ratio decreases very fast. For the motionless case, |v | = 0(m/s), the ratio in all three cases is the same and equal to 1 = 0.7071 while for |v | = 1(m/s), the ratio is R 0.0022 for T = 107 (s) and 0.0071 for T = 106 (s).
7.6
Summary and Conclusion
Throughout this chapter, we have assumed that each user is associated with an ON-OFF process of activity. We have derived analytical expressions for the distribution of energy consumed by a base station. We have found that, with or without mobility, the base station is expected to consume the same amount of energy in average. We have proved that mobility reduced moments of the additive part of energy. We have also proved that high mobility leads the variance of energy to 0. These results are strong since they hold true for any mobility model. In the case of completely aimless mobility model, we have 1 characterized the convergence rate to 0 of the variance, which is |v |.
103
10
T = 105 (s) T = 106 (s) T = 10 (s) 10

1
10
10
10
10
20
30
40
50 |v| (m/s)
60
70
80
90
100
Figure 7.4: Inuence of users speed on the ratio
V[JA ( M ,T )] E[JA ( M ,T )]
104
7. ON-OFF model
105
Chapter 8
Generalized Glauber model

Contents
8.1 8.2 Introduction 8.2.1 8.2.2 8.3 8.3.1 8.3.2 8.3.3 8.4 8.4.1 8.4.2 8.4.3 8.5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105 Model description and main results . . . . . . . . . . . . . . . . 106 Generalized Glauber dynamic . . . . . . . . . . . . . . . . . . . . 106 Generalized Glauber dynamic with mobility . . . . . . . . . . . . 109 Proof of Theorem 48 . . . . . . . . . . . . . . . . . . . . . . . . . 115 Proof of theorem 49 . . . . . . . . . . . . . . . . . . . . . . . . . 116 Proof of theorem 50 . . . . . . . . . . . . . . . . . . . . . . . . . 116 Lemmas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116 Proof of Theorem 53 . . . . . . . . . . . . . . . . . . . . . . . . . 119 Special case: Completely aimless mobility model . . . . . . . . . 120 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
Analysis in no mobility case . . . . . . . . . . . . . . . . . . . . . 110
Impact of mobility . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
Conclusion
8.1
Introduction
The previous chapter considers users presented by a random conguration on d dimension, each user is associated with an ON-OFF process of activity to study the distribution of consumed energy. We consider in this chapter another model called generalized Glauber dynamic. Glauber dynamic can be described as following. It starts at t = 0 with a conguration on a bounded domain D . Each point of this conguration has an exponential life time, after the life time the point disappears. In parallel, there is new arriving points. The arriving time follows a Poisson point processon R+ and points are randomly placed on D according to some distribution. Glauber dynamic has been already successfully applied to study the blocking rate of cellular network. For this it requires that the sojourn time is exponentially distributed so that the system can be modeled as a Markov process that takes value in the space of nite point measures. The author of [56] call it spatial Markov queueing process. In queueing theory, under many situations it can be argued that the blocking rate does not depend on the distribution of sojourn time but only on its mean. This property is well known as insensibility property. On the contrary, the distribution of
106 s
8. Generalized Glauber model
t0
Figure 8.1: Generalized Glauber spatial dynamic, similar to a M/G/ queue a user is characterized by an arrival time, a sojourn duration, however in our model he is also characterized by his position and his mobility process. Consequently there may be innite users at a time t but the number of users in cell C is always nite.
energy consumed during certain duration depends heavily on the distribution of sojourn time, as showed latter in this chapter. We consider in this chapter a more general scenario and so we call it generalized Glauber dynamic because of the following arguments. Firstly the domain we consider is no longer bounded but it can be, for instance, Rd so the . Secondly the system starts at . Thirdly we no longer assume that the sojourn time is exponential distributed but we only assume that the sojourn time is positive and has the nite mean. Furthermore we use the Poisson point process approach instead of Markov process approach. This chapter is organized as follows. Section 8.2 describes the model and presents main results. Section 8.3 presents proofs in the no mobility case. In the section 8.4 the impact of mobility is treated.
8.2
8.2.1
Model description and main results

Generalized Glauber dynamic
Let be a nite Radon measure on Rd , absolutely continuous with respect to the Lebesgue measure with density p . Let L,S be the Poisson point process of intensity
107 measure given by d (t, s, x) = dt pS (s) ds d(x)
where pS (s) is a probability density function of a positive random variable S with nite mean. A realization of this process, if (Rd ) < , can be obtained as follows. Let NA = (Ti , i 0) be a Poisson point process of intensity (Rd ) on R and ((Xi , Si ), i 0) be a sequence of i.i.d random variable, independent of NA , such that dPXi (x) = d(x) and dFSi (s) = pS (s) ds (Rd )
Then L,S = i0 (Ti ,Xi ,Si ) . In particular, for a bounded domain C , one can choose as the Lebesgue measure restricted on C . If (Rd ) = , since is nite, there exists a sequence of compact sets (Kk , k 1) such that (Kk ) and k Kk = Rd , then L,S is the weak limit of the sequence (Lk ,S , k 1) where k is the restricted of to Kk . The conguration of active users at time t is Nt =
i 1
1{Ti t<Ti +Si } Xi
Roughly speaking, in this model a user i arrives at time Ti at location Xi and stays in the system during Si units of time. From denitions in the chapter 6, for this conguration model, it follows that the total energy consumed between 0 and T is given by: JT (N, T ) =
i 0
1{Ti T } (((Ti + Si )+ T ) Ti+ )(Xi ) +
( Nt ) dt
0
= JA (N, T ) + JB (N, T ) Theorem 48. Assume that (Rd ) < , the expectation of JT (N, T ) is given by: E [JT (N, T )] = T E [S ] where p
N0
(x) d(x) + T
Rd 0
N0
(r ) (r ) dr,
(8.1)
is the probability density function of N0 and is given by: p

N0
(r ) =
E [S ] d(C (r )) E[S ](C (r)) e dr

2
(8.2)
The second order moment of JT (N, T ) is given by: m2 [JT (N, T )] = T

Rd
(x) d(x)
+ K2 (T )
2 (x) d(x)
Rd
+LHS of (8.12) + 2 LHS of (8.10) + 2 LHS of (8.13) where

T 0
Kn (T ) =
T
dt E
(t + s)+ T t+
n
pS (s) ds
(t + S )+ T t+
dt.
108
In order to gain some insights on the behavior of JT (N, T ), we use the concentration inequality for Poisson point process to obtain the following result: Theorem 49. Assume that S S (a.s) and (x) + (x) B for any x Rd with S , B > 0. Let (T ) = (S T )B and 2 (T ) = K2 (T ) then P(JT (N, T ) ) exp where g(u) = (1 + u) ln(1 + u) u. If we count only the number of users in the cell C , the system can be seen as a M/G/ queue. The returning time to empty state of this queue always has nite mean. Consequently, the time varying conguration process N is a regenerative process since it can be split into i.i.d cycles. It is well known that under appropriate conditions, for a T general model of power consumption PG , the consumed energy JG (N, T ) = 0 PG (Nt ) dt can be well approximated by a Gaussian random variable as T . Unfortunately, we can only nd an error bound on Gaussian approximation of JA (N, T ), as follows: Theorem 50. We have: P (JA (N, T ) u) Q for any u > 0. As showed later in Lemma 59, Kn (T ) T mn [S ], thus the bound becomes m3 [S ] T m2 [S ]
Rd Rd
((x) + (x))2 d(x)

Rd
(T ) (T )
g (
L E [JT (N, T )]) (T ) 2 (T )
(8.3)
u E [JA (N, T )] V [JA (N, T )]
K3 (T ) Rd 3 (x) d(x) (K2 (T ) Rd 2 (x) d(x))3/2
3 (x) d(x) 2 (x) d(x)

3/2
when T is large. In contrary to the ON-OFF model without mobility where the bound 1 decays as O(1), in this model the bound decays as O( ) as T . T Now, if d(x) = dx then the bound becomes:
1 which decays as O( ).
K3 (T ) (K2 (T )
Rd Rd
3 (x) dx 2 (x) dx)3/2
109
8.2.2
Generalized Glauber dynamic with mobility
From now on, when considering mobility, we will make the assumption that d(x) = dx. The model is the same as the above Generalized Glauber dynamicmodel, except that each user moves during its sojourn. The mobility model is already described in the chapter 6. Let M1 (t), M2 (t), ... be independent versions of M (t) and be the movement of users 1, 2, ... respectively. In other words, the position of user i during at time t [Ti , Ti + Si ) is Xi + M (t Ti ). The marked point process L dx,S,M = ((Ti , Si , Xi , (Mi (t), t R))) i=1 is Poisson point process on R R Rd D ([0, ), Rd ) due to the independent marking property. Its measure intensity is: d (t, s, x, (m(t), t R)) = dt pS (s) ds d(x) d PM (m) At time t, the conguration of active users is NM =
i 1
1{Ti t<Ti +Si } Xi +Mi (tTi )
Lemma 51. N M is a Poisson point process of intensity measure E [S ] dx for all t. The proof of the above lemma is given in the subsection 8.4.1. From the above lemma, we immediately deduce that: Theorem 52. For any power consumption model PG and mobility model M , we have E JG (N M , T ) = E [JG (N, T )]. In particular E JA (N M , T ) = E [JA (N, T )], E JB (N M , T ) = E [JB (N, T )] and E JT (N M , T ) = E [JT (N, T )]. We obtain the same result as in the previous chapter. In fact, mobility makes the same impact on the consumed energy in both model, as we will investigate now. Theorem 53. We have, E exp JA N M , T for any R and cn JA N M , T mn JA N , T
M
E [exp {JA (N, T )}]
mn [JA (N, T )] .
n
cn [JA (N, T )]
for all n 2. If M satises the property T then cn JA N M/ , T and V [JB ( N M/ , T )] , V JT N M/ , T 0 as 0. 0 and mn JA N M/ , T E [S ]

Rd
(x) dx
Furthermore, if (x) = a1{xC } (a > 0) and M satises the property T then cn JA N M/ , T and mn JA N M/ , T are decreasing functions of . A bound for Gaussian approximation of JA (N M , T ) is found as follows
110 Theorem 54. We have P JA (N M , T ) u Q
u E JA (N M , T ) V [JA (N M , T )]
M (, T ) 2 3/2 (M 3 (, T ))
for any u > 0. Here M n (, T ) is dened in equation (8.14). When T is large, the bound becomes:
M (, S ) E H2 M (, S ))3/2 T E H3
Using the concentration inequality (corollary 2) we obtain the following bound for the distribution of JT (N M , T ). The proof is similar to that of theorem 49 so we omit it here. Theorem 55. Assume that S S (a.s) and (x) + (x) B for any x Rd with S , B > 0. Let (T ) = (S T )B then P(JT (N M , T ) u + E JT (N M , T ) ) exp M 2 ( + , T ) (T )
2
u (T ) M 2 ( + , T ) (8.4)
where g(u) = (1 + u) ln(1 + u) u.
8.3
Analysis in no mobility case
In this section we present calculations and proof in the case where users do not move during their sojourn. Let Se be the random variable with the associated stationary-excess (or equilibrium-residual-lifetime) CDF ([57], [58]) F Se (s) =
s F S (s1 )
ds1
E [S ]
Its PDF,MGF and nth moments are given by pSe (s) = F S (s) , E [S ]
E etSe =
1 E etS , tE [S ] mn+1 [S ] (n + 1)E [S ]
mn [Se ] =
It is easily seen that if S Exponential() then Se follows the same distribution. We also note that if S u a.s. then Se u a.s. .
111 s
s0 0 t t0
Figure 8.2: Illustration of the proof of lemma 56.
Lemma 56. For any t0 R, Nt is a Poisson point process on Rd with intensity measure E [S ] d(x). Given a realization of Nt , the residual sojourn time of users are independent and follow the same distribution as that of Se . Moreover, the probability distribution function of Nt is given by: P( Nt > r ) = eE[S ](C (r))) and its probability density function is given by: p
Nt
(8.5)
(r ) =
E [S ] d(C (r )) E[S ](C (r))) e dr
(8.6)
Proof. As above dened, the conguration of active users at t, associated with their residual service time is 1{Ti t0 <Ti +Si } (Xi ,Si (t0 Ti ))
i 1
By displacement theorem, this is a Poisson point processwith intensity measure characterized by intensity (D (s0 , )) = = =
0
dt
0
pS (s)1{tt0 ,s>s0 +tt0 } ds
d(x)
D
t0 0
F S (s0 + t t0 ) dt(D ) F S (s0 + t) dt(D )
= E [S ] (D )F Se (s0 ) for all measurable set D Rd and s0 0. This is exactly the same intensity of a independently marked Poisson point process with underlying intensity measure E [S ] d(x) and the distribution of marks are the same as that of Se . Thus, the lemma is proved.
112 s II I
III
t1
t2
Figure 8.3: Illustration of the proof of lemma 57.
Remark 3. For a domain D such that (D ) < , the dynamic of the number of users on D follows exactly the same as the dynamic of the M/G/ queue with arrival rate (D ) and service time distribution dFS . Thus, the model can be called spatial M/G/ queue. Lemma 57. Nt1 Nt2 , Nt1 Nt2 , Nt2 Nt1 are 3 independent Poisson point processes of intensity measure FSe (|t2 t1 |)E [S ] d(x), F Se (|t2 t1 |)E [S ] d(x) and FSe (|t2 t1 |)E [S ] d(x) on Rd , respectively. Proof. (see gure ??) We note that, a user i is in Nt1 Nt2 , Nt1 Nt2 , Nt2 Nt1 if and only if the point (Ti , Si ) is in the domain I,II,III respectively. Thus, by thinning property, Nt1 Nt2 is Poisson point process of intensity characterized by:
t1
intensity (D ) = (D ) = (D )
0
dt
t2 t t1 t
pS (s) ds
(F S (t) F S (t + t2 t1 )) dt
= E [S ] (D ) 1
F S (t + t2 t1 ) dt E [S ] 0 = E [S ] (D )FSe (t2 t1 )
for all measurable set D Rd . Therefore Nt1 Nt2 is a Poisson point process with intensity E [S ] F Se (t2 t1 ) d(x). Similarly Nt1 Nt2 , Nt2 Nt1 are 2 Poisson point processes of intensity measure FSe (t2 t1 )E [S ] d(x) and F Se (t2 t1 )E [S ] d(x) on Rd , respectively. The independence between the three Poisson point process follows that fact that I,II and III are pairwise disjoint.
From the above lemma, we can see that the time varying process N is time reversible. It can be deduced from the time reversibility of a M/G/ queue. Moreover, if S u a.s. then Se u a.s. and Nt1 and Nt2 are independent for |t2 t1 | > u.
113 Theorem 58. For n 1 we have: mn [JA (N, T )] = Bn K1 (T ) and cn [JA (N, T )] = Bn 0, K2 (T ) In particular, E [JA (N, T )] = T E [S ] (x) d(x).
Rd Rd Rd
(x) d(x), ..., Kn (T )
n (x) d(x)
Rd
2 (x) d(x), ..., Kn (T )
n (x) d(x)
Rd
(8.7)
Proof. As JA (N, T ) is a linear functional of L,S , we can apply the theorem 7.4. It is sucient to note that:
R R R d
(t + s)+ T t+ (x) = Kn (T )
Rd
pS (s) dt ds d(x)
n (x) d(x).
Lemma 59. We have the following properties: 1. Kn (T ) = E (S T )n (S T ) Proof. 1) We have: Kn (T ) =

0 0 0 T n 1 n+1 (S
T )n+1 .
2. Assume that mn [S ] < then Kn (T ) T mn [S ] as T .
pS (s) ds
0
(t + s)+ T t+
T
dt
n
= =
pS (s) ds
0
+
0
(t + s)+ T ) t+
T 0
dt
pS (s) ds
s
((t + s)+ T )n dt +
(((t + s) T ) t)n dt
By some simple manipulations we get:

0 s
((t + s)+ T )n dt +
T 0
(((t + s) T ) t)n dt
sn T T ns
n1 n+1 , n+1 s n1 n+1 , n+1 T
s T; T s.
Thus, 1) is proved. 2) We have Kn (T ) T (S T )n (S T ) = E T E [S n ] and limT (S T )T (S T ) n+1 T (T ) we have limT Kn = mn [S ]. T

n n1
n 1 n+1 (S
T )n+1 T
(S T )n+1
= S n . Thus, by dominated convergence theorem
114
We see that, from the above lemma, when T goes larger, the "border eect" is negligible. Furthermore, we have cn [JA (N, T )] T mn [S ]
Rd
n (x) d(x) (E [S ])n
n (x) d(x).
Rd
(8.8)
That is to say, when T is large, among all positive random variable S having the same expectation, the constant one minimizes the moments of JA (N, T ). Lemma 60. The joint distribution of ( Nt1 , Nt2 ) is given as: F(
Nt 1 , N t 2 E[S ]FSe (|t2 t1 |)(C (u1 )) ) (u1 , u2 ) = 1 e
1 eE[S ]FSe (|t2 t1 |)(C (u2 )) (8.9)
1 eE[S ]F Se (|t2 t1 |)(C (u1 u2 )) for all t1 , t2 R and u1 , u2 > 0. The expectation of JB (N, T ) is given by: E [JB (N, T )] = T
0
(r ) dF
N0
(r )
and its second order moment is given as:

T T
m2 [JB (N, T )] =
0 0
dt1 dt2
0
(u1 ) (u2 ) dF(
Nt 1 , N t 2
) (u1 , u2 ).
(8.10)
Proof. We have F(
Nt 1 , Nt 2
= 1 eE[S ]FSe (|t2 t1 |)(C (u1 ))
= P( Nt1 Nt2 u1 )P( Nt2 Nt1 u2 )P( Nt1 Nt2 u1 u2 ) 1 eE[S ]FSe (|t2 t1 |)(C (u2 )) 1 eE[S ]F Se (|t2 t1 |)(C (u1 u2 ))
) (u1 , u2 ) = P( Nt1 u1 , Nt2 u2 ) = P ( Nt1 Nt2 u1 , Nt2 Nt1 u2 , Nt1 Nt2 u1 u2 )
as Nt1 Nt2 , Nt2 Nt1 , Nt1 Nt2 are three independent Poisson point processes (lemma 57). Using Fubinis theorem the expectation of JB (N, T ) is expressed as:
T
E [JB (N, T )] =
0
E ( Nt ) dt
0
=T
(r ) dF
N0
(r )
as Nt has the same distribution for all t. Similarly, the second order moment of JB (N, T ) is:
T 2
m2 [JB (N, T )] = E
0 T T 0 T T
( Nt ) d t E ( Nt1 ) ( Nt2 ) dt1 dt2 dt1 dt2

0 0
=
0
=
0 0
(u1 ) (u2 ) dF(
Nt 1 , Nt 2
) (u1 , u2 ).
115 It is worth noting that we can generalize result of lemma 57 to calculate nth order moment of JB (N, T ). For example, for n = 3 and t1 < t2 < t3 one can prove that Nt3 Nt2 , Nt3 Nt2 Nt2 , Nt2 Nt1 Nt3 , Nt3 Nt1 , Nt2 Nt1 Nt3 , Nt1 Nt2 are six independent Poisson point processes on Rd .
8.3.1
Proof of Theorem 48
In this subsection we prove the results presented in theorem 48. As JT (N, T ) = JA (N, T ) + JB (N, T ) we apply lemmas 58 and 60 we obtain the analytical expression of E [JT (N, T )]. We are now interested in m2 [JT (N, T )]. We have m2 [JT (N, T )] = m2 [JA (N, T )] + 2E [JA (N, T )JB (N, T )] + m2 [JB (N, T )] . (8.11)
Expressions for m2 [JA (N, T )] and m2 [JB (N, T )] have been found in the lemmas 58 and 60 so it remains to calculate E [JA (N, T )JB (N, T )]. Using Fubinis theorem the later can be written as:
T T
E [JA (N, T )JB (N, T )] = E

0 T
PA (Nt1 ) dt
T
( Nt2 ) dt2
=
0 T 0 T 0 T 0
E PA (Nt1 ) ( Nt2 ) dt1 dt2 E PA (Nt1 Nt2 ) ( Nt2 ) dt1 dt2

T 0
=
0
E PA (Nt1 Nt2 ) ( Nt2 Nt1 ( Nt2 Nt1 ) dt1 dt2
= term1 + term2 Recall from lemma 57 that Nt1 Nt2 is independent of Nt2 so term1 = E [S ] = 2E [S ]
Rd
(x) d(x)
Rd 0
T 0 T
(r ) dF
N0
(r )
0
FSe (|t2 t1 |) dt1 dt2 (8.12)
(x) d(x)
0
(r ) dF
N0
(r )
0
(T t)FSe (t) dt
Now, also from 57, the joint distribution of ( Nt1 Nt2 , Nt2 Nt1 ) is given as: F(
Nt1 Nt2 , Nt2 Nt1
) (u1 , u2 ) = 1 eE[S ]F Se (|t2 t1 |)(C (u2 u2 )) .
1 eE[S ]FSe (|t2 t1 |)(C (u1 ))
Note that conditioning on the event " Nt1 Nt2 = u1 ", Nt1 Nt2 is a Poisson point process of intensity E [S ] F Se (|t2 t1 |) on C (u1 ). Thus, by Campbells theorem, we have:
T T 0
term2 = E [S ]
0 0 0
F Se (|t2 t1 |) dt1 dt2 (x) dx (u1 u2 ) dF(

Nt1 Nt2 , Nt2 Nt1
C (u1 )
) (u1 , u2 ).
(8.13)
116
8.3.2
Proof of theorem 49
We now have give a proof for theorem 49. Note that JT (N, T ) is a functional of L,S and: D(t,s,x) JT (N, T ) (((t + s)+ T ) t+ )((x) + (|x|)) (S T )B Then we apply the theorem 1 we obtain the desired result.
8.3.3
Proof of theorem 50
The result is consequence of 10 as JA (N, T ) is a linear functional of L,S .
8.4
Impact of mobility
In this section we consider the impact of mobility. In this case, the consumed energy can be written as JT (N , T ) =
i 0 0 M ((Ti +Si )+ T )(Ti+ T ) T
(Xi + Mi (u)) du +
0
( NtM ) dt
= JA (N M , T ) + JB (N M , T ) Consequently, JA (N M , T ) is a linear functional of L dx,M,S .
8.4.1
Lemmas
Proof. (of lemma 51) It is sucient to prove that for all t, ((Ti , Xi + M (t Ti ))) i=1 is a Poisson point , process with the same intensity measure as that of ((Ti , Xi )) i=1 i.e dt dx. To prove that, we apply the displacement theorem. Let l1 , ld be the Lebesgue measures on R and Rd . For any subsets A R and B Rd , applying the displacement theorem we have d that ((Ti , Xi + M (t Ti ))) i=1 is a Poisson point process on R R with intensity measure given by: (A B ) = =
A
du
A Rd
P(M (t u) + x B ) dx
Rd
du
Rd
pM (tu) (y ) dy 1{y+xB } dx
Rd
=
A
du
Rd
pM (tu) (y ) dy
Rd
1{y+xB } dx
= ld (B )
A
du
pM (tu) (y ) dy
= l1 (A)ld (B ) We conclude the proof. Let M n (f, T ) =

T
dt
0
M pS (s)Hn (f, (t + s)+ T t+ ) ds
(8.14)
117
M (f, u) is dened in with u R+ and f : Rd R+ such that f (x) = 0 in Rd /C and Hn equation (7.10).
M n (f, T ) =
T
dt
0
pS (s) ds
0
(t+s)+ T t+
...
0
(t+s)+ T t+
M n (f, u1 , ..., un ) du1 du2 ... dun .
If S = constant then: M n (f, T ) = 2 2

T 0 S 0 M (f, u) du + (S T )H M (f, T ), S T ; Hn n M (f, u) du + (T S )H M (f, S ), S T . Hn n
(8.15)
M We see that in this case when T then M n (f, T ) T Hn (f, S ). In the lemma 62 we generalize this in the case where S is random. n n M M (f, u) un Lemma 61. We have Hn Rd f (x) dx, and n (f, T ) Kn (T ) Rd f (x) dx for all n 1, u [0, ). Equality occurs if n = 1. Moreover, if M has the property T then: M/ (f, u) = lim M/ lim Hn n (f, T ) = 0. 0 M/
If f (x) = a1{xC } then M n (f, u) and Hn and a > 0.
(f, u) are decreasing function of with n 2
Proof. We have Rd f n (x + y ) dx = Rd f n (x) dx for any y Rd . Now for n 2, for any y1 , ..., yn Rd , apply the CauchySchwarz inequality we have:
n Rd i=1 n
f (x + yi ) dx =
i=1
Rd
f (x + yi ) dx
1 n
f n (x) dx.
Rd
Therefore,
n
M n (f, u1 , u2 , ..., un ) = E and

M Hn (f, u) = u u
Rd i=1
f (x + M (ui )) dx
f n (x) dx.
Rd
...
0 Rd 0 0
M n (f, u1 , ..., un ) du1 du2 ... dun
un M n (f, u)
T
f n (x) dx.
dt
((t + s)+ T t+ )n pS (s) ds
f n (x) dx
Rd
= Kn (T )
f n (x) dx.
Rd
118
M (u2 ) (u1 Now as M (u1 ) as 0 then for suciently small one of two point x+ M M (ui ) and x + M (u2 ) must be outside the support of f . Consequently n i=1 f (x + ) 0 as 0. By dominated convergence theorem we obtain lim0 M/ (f, u1 , u2 , ..., un ) = 0 and we have M/ lim Hn (f, u) = 0.
Using (8.14) and applying the dominated convergence theorem we have:

0
lim M/ n (f, u) = 0.
M/
Finally, if n 2 and f (x) = a1{xC } with a > 0 then Hn decreasing functions because M/ (f, u1 , ..., un ) is.
(f, u) and n
M/
(f, u) are
M Lemma 62. We have M n (f, T ) T E Hn (, S ) as T .
Proof. We have: M n (f, T ) T = 1 T +

0
dt
T 0
M Hn (f, (t + s)+ T t+ )pS (s) ds
1 M Hn (f, (t + s) T t)pS (s) ds dt T 0 0 = term1 + term2 .
It is easy seen that term1 0 as T . We then consider term2 . We can write

1
term2 =
0
M E Hn (f, (tT + S ) T tT ) dt
M (f, s) is an increasing Since (T t + s) T tT s, limT ((T t + S ) T tT ) = s and Hn function of s, by monotone convergence theorem we have: M lim term2 = E Hn (f, S ) .
Theorem 63. The moments and central moments of JA (N M , T ) are given as: cn JA (N M , T ) mn JA (N M , T ) for all positive integer n. Proof. The result follows the linearity property of JA (N M , T ) and the theorem 7.4 and we note that:
((t+s)+ T )(t+ T ) RRRd D (R,Rd ) 0 n M = Bn (0, M 2 (, T ), ..., n (, T )) M M = Bn (M 1 (, T ), 2 (, T ), ..., n (, T ))
(x + M (u)) du
dtpS (s) ds dx dPM
= M n (, T ).
M (, S ) . When T is large, use lemma 62 we can write: V JA (N M , T ) T E H2
119
8.4.2
Proof of Theorem 53
With the assumption d(x) = dx, applying theorem 63 and lemma 61 we have: mn [JA (N, T )] = Bn K1 (T ) (x) dx, ..., Kn (T ) n (x) dx
Rd
Rd
M B n M 1 (, T ), ..., n (, T )
= mn JA (N M , T )
and cn [JA (N, T )] = Bn 0, K2 (T ) 2 (x) d(x), ..., Kn (T ) n (x) dx

Rd
Rd
M Bn 0, M 2 (, T ), ..., n (, T )
= cn JA (N M , T ) .
Now we have, as 0:
n
mn JA (N M/ , T ) Bn (T and
(x) dx, 0..., 0) =

Rd
T
Rd
(x) dx
cn JA (N M/ , T ) Bn (0, ..., 0) = 0. Now, using the property of linear functional of Poisson point process and Jensens inequality, we have E exp JA (N , T )
T M T
= exp
dt
0
pS (s) ds
e
Rd
(t+s)+ T t+
(x+M (u)) du
dx
+ T )t+ )
exp = exp
T
dt
T 0
pS (s) ds pS (s) ds
dx
Rd
1 + ((t + s) T ) t+
t+
(t+s)+ T t+
e(x+M (u))(((t+s)
du
dt
0 0
1 + ((t + s) T ) t+ e(x)((t+s)
(t+s)+ T
du
Rd
e(x)(((t+s)
+ T )t+ )
exp
dt
pS (s) ds
+ T )t+
Rd
= E [exp {JA (N, T )}] . Here we use convention Now we have:

1 0 t ... t e
du = 1. 1
M/ M/
0 V JB (N M , T ) Apply lemma 61 we have

0
0 V JT (N M , T )
( + , T ) (, T ).
lim V JT (N M/ , T ) = lim V JB (N M/ , T ) = 0.
0
120
8.4.3
Special case: Completely aimless mobility model
We consider now the complete aimless mobility model with constant speed, i.e M (t) = tv where |v | = constant and the direction of v is uniformly distributed. Lemma 64. Consider the completely aimless mobility model then: 1. If f (x) = a1{xC } then
M Hn (f, u) = O
1 |v |n1
, |v |
M (f, u) = O (u), u for n 2. and Hn
2. If d = 1 and C = [R, R] and f (x) = a1{|x|R} with a > 0 then: we have M n (f, T ) = E [h(S T, T S )] where: h(s, t) = t 2Rsn t (8.16)
n1 n+1 |v | n+1 s n+1 (2 R ) (2R)n s n 1 n +1 |v|n |v|n1
an
n(n1) n 1 n+1 + n+1 2Rs (n+1)(n+2) 1 (2R)n+1 s n an n a , n+1 |v |n
|v | sn+2 an , s s
2R |v | ; 2R |v | .
Proof. 1) can be proved following the same lines as the proof of lemma 45 in the previous chapter, thus we omit here. Now we prove 2) . Using equations (8.14) (8.15) we need only prove (8.16) in the case R where S = constant < T . In this case if S 2 |v| then: M n (f, T ) = 2 n 1 n+1 n 1 n+1 u |v | du + (T S ) 2RS n S |v | n+1 n+1 0 n1 n(n 1) n 1 n+1 S |v | 2RS n+1 + |v | S n+2 = T 2RS n n+1 n+1 (n + 1)(n + 2) 2Run
S
and if S
2R |v | :
2R |v |
M n (f, T ) = 2
0 S
2Run (2R)n u |v |n1
n 1 n+1 u |v | n+1
du du + (T S ) (2R)n S |v |n1 n 1 (2R)n+1 n + 1 |v |n
+2 = T
2R |v |
(2R)n S n 1 (2R)n+1 n + 1 |v |n |v |n1
n 1 (2R)n+1 n + 1 |v |n
n(n 1) (2R)n+2 n 1 (2R)n+1 S + . n+1 |v |n (n + 1)(n + 2) |v |n+1
This completes the proof. Applying the above results, we have expressions in the special case: d = 1, C = [R, R], (x) = 1{|x|R} , (x) = b1{|x|R} and M (t) = tv with v = |v | or |v | with equal probability 1 2 . In particular, if S = constant < T we have: E JT (N M , T ) = E JA (N M , T ) + E JB (N M , T ) = 2T RS + T b(1 e2SR ),
121
3
2 1.8 1.6 1.4 1.2 1 0.8 0.6 0.4 0.2
x 10
Exponential Uniform Deterministic
10
20
30
40
50 |v| (m/s)
60
70
80
90
100
Figure 8.4: Impact of distribution of sojourn time and users speed on the motionless case.
V[JA (N M ,T )] JA (N,T )
in the
V JA (N M , T ) = T if S
2R |v |
2RS 2
|v | S 3 3
2RS 3 |v | S 4 + 3 6
and V JA (N M , T ) = T (2R)2 S (2R)3 |v | 3 |v |2 (2R)3 S 3 |v |2 + (2R)4 6 |v |3

2R |v|
if S
and the high mobility regime corresponds to S M n (, T )

T E[S ](2R)n |v|n1
2R |v | .
We can think that, in this case, the low mobility regime corresponds to S . In particular,
2R |v | .
If S is random and |v | , then
V JA (N M , T )
T E [S ] (2R)2 |v |
A in function of is the same as 360(s) = 6(mins). Figure 8.4 plots the ratio JA (N,T ) |v |. Since (x) = 1{|x|R} , the ratio is decreasing function of |v |, as expected. The variance is smallest in the case of deterministic sojourn time, which seems intuitive. We observe that for |v | 3(m/s) the ratios are almost the same in three distributions of sojourn time while they dier enormously in small values of users speed. Figure 8.5 plots the PDF of V [JA (N, T )] (motionless users). We see that even in this simple case, there is dierence between the distribution of JA (N, T ). The variance of JA (N, T ) in the case of deterministic is smallest as predicted in equation 8.8 because T is large.
For a numerical application, we take d = 1, 2R = 1000(m), T = 107 (s) 115(days), = 5.55555556.105 (users/(m.s)). We also choose (x) = 1{|x|R} so that we have exact expressions for JA . For the distribution of sojourn time, we consider the three cases: S = 360(s) (deterministic), S exponential(360) (exponential distributed) and S unif orm(0, 720) (uniformly distributed). In all three cases, the average sojourn time
V[J (N M ,T )]
122
6
1.5
x 10
Exponential Uniform Deterministic
0.5
0 1.98
1.985
1.99
1.995
2.005
2.01
2.015
2.02 x 10
8
Figure 8.5: Impact of distribution of sojourn time on the distribution of JA (N, T ) in the motionless case.
8.5
Conclusion
In this chapter we have presented a model called generalized Glauber dynamic to evaluate the consumed energy in a base station of cellular networks between 0 and T where user arrives according to a Poisson point process in time-space and they are associated with a random sojourn time and a mobility process. For the additive part of consumed energy JA , the moments can be expressed in term of Bells polynomial for both cases: motionless and mobile users. We have also determined an upper bound for the distribution of JA and that of the Gaussian one with the same mean and variance. In the case of no mobility, we have found an ideal to calculate the moments of JT but it is too complex for n 3 so we have done it for the 2th moment. We have found asymptotic approximations of the moments of JA when T is large. We have determined the eect of mobility of users. In particular we have showed that it reduces the moments of JA and in high mobility regime, it reduced the variance of JT , JA to 0.
123
Part IV
Conclusion
125
Chapter 9
Conclusion and future works

Contents
9.1 9.2 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125 Future works . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
9.1
Summary
In this thesis we explored tools of stochastic analysis and Poisson point process applied to cellular networks. We rst introduced some basics on Poisson point process and Malliavin calculus. In the rst application, we found a relationship between the probability of overloading the system, the density of active users and the number of available subcarriers, thus providing a large number of possibilities to design OFDMA systems. We calculated the probability of losing a user in a OFDMA system because all subcarriers of the base station are already in use by several methods including Gaussian approximation, Edgeworths expansion and upper bound derived from the concentration inequality. We apply to dimension the number of subcarriers so that the loss probability is small. We also compared the numerical results with simulations and note that the upper bound of overloading probability leads to an overestimate of the number of subcarriers by about 15% of the simulated one. We also compared it to Gaussian approximation and Edgeworths expansion and found that it is more robust against uncertainty on systems parameters. The margin provided by the bounds may be viewed as a protection against errors in the modeling or in the estimations of parameters. In the second application, we developed energy consumption models for cellular networks. We rst dened the power consumption model for base station as a function of the collection of positions of users. The consumed power consists of the power dedicated to broadcast the same information to all active users in the cell and the power dedicated to transmit, receive, decode and encode the signal of any active user. The total consumed energy during a period is the sum of all consumed power, which in turn is divided into the broadcast part and the additive part. We consider the rst model where at each instance, the conguration of active users follows a homogenous Poisson point process. In this model, we proved that the consumed energy is a increasing function of cell radius. Taking into account the cost of base stations, we proved that there exists an optimal cell radius in the economical point of view. We also present the mobility model for users where each user moves independently from others but statistically identical .
126
9. Conclusion and future works
We then dene two models for users activity. In the rst model, each user is associated with an on-o process and the system serves only on users. In the second model, called generalized Glauber model, calls arrive following a Poisson point process being in the system for some time period and then disappear. The key idea is that the system can be described by a Poisson point process, so we can apply theoretical results. In both models, we were able to nd analytical expressions for the statistics of the consumed energy as well as bounds on its distribution. We showed that the additive part can be approximated by Gaussian random variable and we found an error bound. We then considered the impact of mobility and again we can provide some analytical expressions. Mobility is known to improve the performance of networks in various aspects and we showed that somewhat it has a positive impact on the energy consumption. It turns out that, in both models, with or without mobility the network consumes the same amount of energy in average. However mobility decreases the moments of the additive part and high mobility decreases the variance of consumed energy to zero. In the rst model, we found the decay rate 1 of variance in function of users speed which is O( |v | ). Nevertheless , there are tons of questions that we have not been able to answer yet. We nally remark that, the mathematical frameworks developed to analyze the energy consumption in cellular networks in this thesis can be applied in other studies of wireless system or in queueing theory.
9.2
Future works
In this thesis we employed abstract models to analyze the energy consumption on cellular networks which simplies hypotheses on trac, mobility,... Our energy consumption model is only at the initial state and we need a lot of improvement to model more accurately. Still, we have derived from these abstract models a large number of interesting results that deserve to be further pursued. There are some directions for future research as bellow: 1. Inhomogeneous Poisson process or Non-Poisson process of arrivals, nonPoisson process of positions: In this thesis we assumed a homogeneous Poisson arrival of users (or calls) in the Generalized Glauber model. This means that we assumed that the arrival is always at peak rate. The models can be generalized to inhomogeneous Poisson arrival or periodic inhomogeneous Poisson arrival to reect the fact that in a real network, the arrival of calls is non homogenous or periodic. For example there is always less of calls during the night time than in day time. Moreover, we assumed a Poisson process of positions of users at each instance to take advantage of its randomness and independence aspects. For this reason, we can also expect nonPoisson arrival of calls, or non-Poisson process of positions in the future works to make the model more complete. We can imagine a renewal process as calls arrival process. We can also assume non-Poisson point process such as determinantal point process or cluster process point process to model users position. In this case the system still can be modeled as a point process, but no longer a Poisson point process and therefore need a lot of reection. 2. Loss model: In this thesis, we assumed that the system is not limited on transmitted power or resources so there is no loss of call. In a real system like OFDMA one, the resource is always limited so there is always a small fraction of call to be lost, delayed, or interrupted. If the system is designed so that the probability of losing
127 a communication is very small, out no-loss model can be a good approximation. However, we would like to build a model to capture this fact and we want to know, for instance, if the tendency made by mobility is the same as in non-loss model. We note that spatial loss network is already considered in the literature [56],[59],[60],.... It can be used in the subsequent analysis. 3. Considering shadowing, fading, interference: For the sake of tractability, we assumed that channels between users and base stations are not aected by shadowing or fading. It is possible to take into account these elements in a future research as the system can also described as a Poisson point process. Moreover, cellular network is particularly interference limited. Thus, it would be interesting to measure the impact of interference on the energy consumption of cellular networks. 4. Comparing with simulation/test/real data: Our energy consumption models are purely theoretical. Due to the limit of time, we have not build a simulator or test-bed yet. As a matter of fact, in the future we would like to compare the results obtained in this thesis to a numerical simulation, or a real data. One of our conclusion is that the mobility make no impact on the mean of consumed energy but high mobility decreases its variance to zero. It would be interesting to see if this conclusion is accurate enough in a real situation.
128
Conclusion
129
Chapter 10
Appendix
10.1 ON-OFF exponential process: Basic properties
In this section, assume that I (t) is a stationary ON-OFF exponential process where on1 1 periods and o-periods have mean 1 and 0 respectively. Let 1 = By stationarity we have: P(I (t) = 1) = 1 and P(I (t) = 0) = 0 Lemma 65. [61] We have pij (t, s) = j (1 e(0 +1 )(ts) ) + ij e(0 +1 )|ts| . for all i, j {0, 1}. Moreover, let t1 < t2 < ... < tn and i1 , i2 , ..., in {0, 1} then i1 ,i2 ,...,in (t1 , ..., tn ) = P(n j =1 {I (tj ) = ij }) = i1
T n 1 j =1
1 0 and 1 = . 0 + 1 0 + 1
pij+1 ij (tj +1 , tj ).
Let A(T ) = 0 I (t) dt is the total active time of the user. We can obtain its moment generating function, its expectation, its moments, which is important for the analysis in the next sections. Let V 0 0 1 1 v ( ) and 0 v ( ) 0 , 01 ( ) = , v+ ( ) v ( ) v+ ( ) v ( ) 1 1 v ( ) 10 ( ) = , 01 ( ) = v+ ( ) v ( ) v+ ( ) v ( ) 00 ( ) = , R 0 0 0 1 , B (, t) = exp ((V + R)t) (1 + 0 )2 + 40
1 1 0 + 2
130 Lemma 66. [61] For i, j = 0, 1, we have: Bii (, t) = ii ( )ev+ ()t + (1 ii ( ))ev ()t and
1 1
10. Appendix
Bij (, t) = ij ( )ev+ ()t ij ( )ev ()t ,
E e
A(t)
=
i=0 j =0
i Bij (, t)
To nd the moments of A(t), however it is cumbersome to use the above expression of E eA(t) . We have another way to compute the moments of A(t). Lemma 67. [62] We have:
T tn1 t2
mn [A(T )] = n!1
0
d tn
0
p11 (tn , tn1 ) dtn1 ...
p11 (t2 , t1 ) dt1
(10.1)
In particular, E [A(T )] = and m2 [A(T )] = and m3 [A(T )] = 6T 2 2 6T (1 20 )0 1 T 3 3 0 1 0 + 1 eT (0 +1 ) (0 + 1 )3 (0 + 1 )4 (0 + 1 )5 120 1 (1 0 ) 6T 0 2 1 + + eT (0 +1 ) (0 + 1 )6 (0 + 1 )5

n
0 T 0 + 1
T 2 2 2T 0 1 20 1 0 + + e(0 +1 )T 1 (0 + 1 )2 (0 + 1 )3 (0 + 1 )4
Proof. We have:
T
mn [A(T )] = E
0 T
I (t) dt
T
= E
0 T
...
0 T
I (t1 )...I (tn ) dtn ... dt1 E [I (t1 )...I (tn )] dtn ... dt1
=
0 T
...
0 T
=
0
...
0 T tn 0 T tn 0
P(I (t1 ) = 1, ..., I (tn ) = 1) dtn ... dt1

t2
= n!
0 0 t2
...P(I (t1 ) = 1, ..., I (tn ) = 1) dtn ... dt1

T
= n!
0 0
...
0
p11 (tn , tn1 )...p11 (t2 , t1 )1 dtn ... dt1 ,
where we use the Fubinis theorem and exploit the fact that I (t) = 0 or 1 and I (t) is a Markov chain. Thus, (10.1) is proved. To nd expressions of mn [A(T )] for n = 1, 2, 3 it suces to apply (10.1) with some simple manipulations.
131 Note that for large T , V [A(T )]

2 3 2T 0 1 2 3 0T 0T , m2 [A(T )] , m3 [A(T )] (T ) 3 2 (0 + 1 ) (0 + 1 ) (0 + 1 )3
n n 0T (0 +1 )n as T . 0 T n mn [A(T )] ( as T 0 +1 )
In general, we can prove that mn [A(T )] We also note that for small T ,
0.
10.2
Hermite polynomials
Let be the Gaussian probability density function: (x) = exp(x2 /2)/ 2 and the Gaussian measure on R. Hermite polynomials (Hk , k 0) are dened by the recursion formula: dk Hk (x)(x) = (1)k k (x). dx For the sake of completness, we recall that H0 (x) = 1, H1 (x) = x, H2 (x) = x2 1, H3 (x) = x3 3x
k , using integration by parts, we have Thus, for F Cb
H4 (x) = x4 6x2 + 3, H5 (x) = x5 10x3 + 15x.
F (k) (x) d(x) =

R R R 1(; x] (u)(u)
F (x)Hk (x) d(x). du. Then, Q = and
(10.2)
Let Q(x) =
x (u)
du =
1(; x] (u)Hk (u) d(u) = (1)k

x
(k) (u) du = (1)k Q(k) (x) = Hk1 (x)(x). (10.3)
10.3
Ornstein-Uhlenbeck semi-group
To prove the Edgeworth expansion and its error bound, we introduce some notions of 2 (R; R), we consider Gaussian calculus. For F Cb AF (x) = xF (x) F (x), for any x R. The Ornstein-Uhlenbeck semi-group is dened by Pt F (x) =
R
F (et x +
1 e2t y ) d(y ) for any t 0.
The innitesimal generator A and Pt are linked by the following identity F (x) F (y ) d(y ) =
0
APt F (x) dt.
(10.4)
132
10. Appendix
It is well known that for F C k , (x Pt F (x)) is k + 1-times dierentiable and that we have two expressions of the derivatives (see [63]): e(k+1)t (Pt F )(k+1) (x) = 1 e2t F (k) (et x +
R
1 e2t y )y d(y ).
and (Pt F )(k+1) (x) = e(k+1)t Pt F (k) (x). The former equation induces that (Pt F )(k+1)
e(k+1)t F (k) 1 e2t
e(k+1)t |y | d(y ) = 1 e2t R
2 F (k)
133
List of Figures
2.1 2.2 3.1 3.2 3.3 4.1 4.2 4.3 4.4 5.1 Trade-o between tractability and realistic modeling. . . . . . . . . . . . . . 20 A realization of homogenous Poisson point process of intensity = 2.103 and its thinning with p(z ) = 0.2. . . . . . . . . . . . . . . . . . . . . . . . . 22 OFDMA principle : subcarriers are allocated according to the required transmission rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38 Impact of and on the loss probability (Navail = 92, = 0.0001) . . . . . 45 Estimates of Navail as a function of by the dierent methods . . . . . . . . 47 Outage probability vs SINR threshold . . . . . . . . . . . . . . . . . Handover probability vs SINR threshold . . . . . . . . . . . . . . . . Outage probability vs path loss exponent , Poisson model . . . . . . Handover probability vs path loss exponent , Poisson model, n = 3 . . . . . . . . . . . . . . . . 56 57 57 58
5.2 5.3 5.4 5.5 6.1 6.2 7.1 7.2 7.3 7.4 8.1
Triangles represent BS, plus represent MU. Dotted polygons are Voronoi cells induced by BS. A line between a BS and an MU means that the BS serves the MU. A mobile may be not served by the BS closest to it, due to fading. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Comparison of outage probability between propagation models. For lognormal shadowing = 4(dB ), for Rayleigh fading = 1; K = 102 , = 2.8. . Histogram of no . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Tail distribution of So (f3 ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . A typical histogram of So (f3 ) . . . . . . . . . . . . . . . . . . . . . . . . . .
62 66 70 72 73
Power consumption model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78 Basic model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81 Illustration of the model, each user is associated and a mobility process. . . . . . . . . . . . . . . . Illustration of the proof of Theorem 46. . . . . . Illustration of the proof of Lemma 47. . . . . . . Inuence of users speed on the ratio with . . . . . . . . . a . . . ON-OFF process . . . . . . . . . . . 87 . . . . . . . . . . . 100 . . . . . . . . . . . 101
V[JA ( M ,T )] E[JA ( M ,T )]
. . . . . . . . . . . . . . 103
Generalized Glauber spatial dynamic, similar to a M/G/ queue a user is characterized by an arrival time, a sojourn duration, however in our model he is also characterized by his position and his mobility process. Consequently there may be innite users at a time t but the number of users in cell C is always nite. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
134 8.2 8.3 8.4 8.5
LIST OF FIGURES Illustration of the proof of lemma 56. . . . . . . . . . . . . . . . . . . . . . . Illustration of the proof of lemma 57. . . . . . . . . . . . . . . . . . . . . . . M V[JA (N ,T )] Impact of distribution of sojourn time and users speed on the JA (N,T ) in the motionless case. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Impact of distribution of sojourn time on the distribution of JA (N, T ) in the motionless case. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111 112 121 122
135
List of Tables
1.1 1.2 Mathematical notations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 Abbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
136
LIST OF TABLES
137
Bibliography
[1] A. Fehske, G. Fettweis, J. Malmodin, and G. Biczok, The global footprint of mobile communications: The ecological and economic perspective, IEEE Communications Magazine, vol. 49, pp. 5562, Aug. 2011. [2] Z. Hasan, H. Boostanimehr, and V. K. Bhargava, Green cellular networks: A survey, some research issues and challenges, CoRR, vol. abs/1108.5493, 2011. [3] M. K. Karray, Analytical expressions for blocking and dropping probabilities for mobile streaming users in wireless cellular networks, Wirel. Netw., vol. 16, pp. 2281 2296, November 2010. [4] M. K. Karray, Users mobility eect on the performance of wireless cellular networks serving elastic trac, Wirel. Netw., vol. 17, pp. 247262, January 2011. [5] Y. Zhang, L. Li, and B. Li, Mobility inuences on the capacity of wireless cellular networks, ETRI Journal, vol. 28, pp. 173185, December 2006. [6] L. Decreusefond, E. Ferraz, P. Martins, and T.-T. Vu, Robust methods for lte and wimax dimensioning, in Valuetools, (Cargese, France), June 2012. [7] T. T. Vu, L. Decreusefond, and P. Martins, An analytical model for evaluating outage and handover probability of cellular wireless networks, in The 15th International Symposium on Wireless Personal Multimedia Communications (WPMC12), (Taipei, Taiwan), Sept. 2012. [8] F. Baccelli and B. Blaszczyszyn, Stochastic geometry and wireless networks, volume 1: Theory, Foundations and Trends in Networking, vol. 3, no. 3-4, pp. 249449, 2009. [9] N. Privault, Stochastic Analysis in Discrete and Continuous Settings With Normal Martingales, vol. 1982 of Lecture Notes in Mathematics. Springer, 2009. [10] D. J. Daley and D. Vere-Jones, An introduction to the theory of point processes. Vol. I: Elementary theory and methods. Probability and its Applications (New York), New York: Springer, second ed., 2003. [11] D. J. Daley and D. Vere-Jones, An introduction to the theory of point processes. Vol. II: General theory and structure. Probability and its Applications (New York), New York: Springer, second ed., 2008. [12] D. Stoyan, W. Kendall, and J. Mecke, Stochastic geometry and its applications. Wiley series in probability and statistics, 1995.
138
BIBLIOGRAPHY
[13] G. Peccati, J. L. Sol, M. S. Taqqu, and F. Utzet, Steins method and normal approximation of poisson functionals, Annals of Probability, vol. 38, no. 2, p. 32, 2008. [14] L. Wu, A new modied logarithmic Sobolev inequality for Poisson point processes and several applications, Probability Theory and Related Fields, vol. 118, no. 3, pp. 427 438, 2000. [15] L. Decreusefond, A. Joulin, and N. Savy, Upper bounds on Rubinstein distances on conguration spaces and applications, Communications on stochastic analysis, 2010. [16] C. Houdr and N. Privault, Concentration and deviation inequalities in innite dimensions via covariance representations, Bernoulli, vol. 8, no. 6, pp. 697720, 2002. [17] F. Baccelli, M. Klein, M. Lebourges, and S. Zuyev, Stochastic geometry and architecture of communication networks, Telecommunication Systems, vol. 7, pp. 209227, 1997. [18] F. Baccelli, M. Klein, M. Lebourges, and S. Zuyev, Gomtrie alatoire et architecture de rseaux, Annals of Telecommunications, vol. 51, pp. 158179, 1996. [19] F. Baccelli, B. Baszczyszyn, and P. Mhlethaler, An Aloha protocol for multihop mobile wireless networks, IEEE Trans. Inf. Theory, vol. 52, pp. 421436, 2006. [20] F. Baccelli, B. Baszczyszyn, and P. Mhlethaler, Stochastic analysis of spatial and opportunistic Aloha, IEEE JSAC, special issue on Stochastic Geometry and Random Graphs for Wireless Networks, vol. 27, pp. 11051119, 2009. [21] F. Baccelli and B. Baszczyszyn, A new phase transition for local delays in MANETs, in Proc. of IEEE INFOCOM, (San Diego CA), 2010. [22] B. Baszczyszyn and P. Mhlethaler, Stochastic analysis of non-slotted Aloha in wireless ad-hoc networks, in Proc. of IEEE INFOCOM, (San Diego, CA), 2010. [23] A. Hunter, J. Andrews, and S. Weber, Transmission capacity of ad hoc networks with spatial diversity, Wireless Communications, IEEE Transactions on, vol. 7, pp. 5058 5071, december 2008. [24] A. Busson, An overview of results on ad hoc network performances using spatial model, in Computers and Communications, 2009. ISCC 2009. IEEE Symposium on, pp. 36 41, july 2009. [25] M. Haenggi, Outage, local throughput, and capacity of random wireless networks, Trans. Wireless. Comm., vol. 8, pp. 43504359, Aug. 2009. [26] B. Liu, P. Brass, O. Dousse, P. Nain, and D. Towsley, Mobility improves coverage of sensor networks, in Proceedings of the 6th ACM international symposium on Mobile ad hoc networking and computing, MobiHoc 05, (New York, NY, USA), pp. 300308, ACM, 2005. [27] P.-J. Wan and C.-W. Yi, Coverage by randomly deployed wireless sensor networks, IEEE/ACM Trans. Netw., vol. 14, pp. 26582669, June 2006.
139 [28] R. Ganti, F. Baccelli, and J. Andrews, A new way of computing rate in cellular networks, in Communications (ICC), 2011 IEEE International Conference on, pp. 1 5, june 2011. [29] H. Dhillon, R. Ganti, F. Baccelli, and J. Andrews, Coverage and ergodic rate in k-tier downlink heterogeneous cellular networks, in Communication, Control, and Computing (Allerton), 2011 49th Annual Allerton Conference on, pp. 1627 1632, sept. 2011. [30] E. Ferraz, Algebraic Topology of Random Simplicial Complexes and Applications to Sensor Networks. PhD thesis, Telecom Paristech, 2011. [31] S. Janson, Bounds on the distributions of extremal values of a scanning process, Stochastic Processes and their Applications, vol. 18, no. 2, pp. 313328, 1984. [32] L. Decreusefond, Stochastic Modeling and Analysis of Telecoms Networks. John Wiley & Sons, 2012. [33] R. Agarwal, V. R. Majjigi, Z. Han, R. Vannithamby, and J. M. Cio, Low complexity resource allocation with opportunistic feedback over downlink ofdma networks, IEEE Journal on Selected Areas in Communications, vol. 26, no. 8, pp. 14621472, 2008. [34] J. Huang, V. Subramanian, R. Agrawal, and R. Berry, Downlink scheduling and resource allocation for ofdm systems, Wireless Communications, IEEE Transactions on, vol. 8, no. 1, pp. 288 296, 2009. [35] K. Seong, M. Mohseni, and J. Cio, Optimal resource allocation for ofdma downlink systems, in Information Theory, 2006 IEEE International Symposium on, pp. 1394 1398, 2006. [36] Z. Shen, J. G. Andrews, and B. L. Evans, Adaptive resource allocation in multiuser OFDM systems with proportional rate constraints, Wireless Communications, IEEE Transactions on, vol. 4, no. 6, pp. 27262737, 2005. [37] R. Giuliano and F. Mazzenga, Dimensioning of ofdm/ofdma-based cellular networks using exponential eective sinr, IEEE Trans. Wireless Commun., vol. 58, pp. 4204 4213, Oct. 2009. [38] M. Maqbool, M. Coupechoux, S. Doirieux, and B. Baynat, Dimensioning methodology for ofdma networks, in Proc. of the Wireless World Research Forum (WWRF22), 2009. [39] B. Baszczyszyn and M. Karray, Dimensioning of the downlink in OFDMA cellular networks via an erlangs loss model, in Proc. of European Wireless Conference, (Aalborg), 2009. [40] M. K. Karray, Analytical evaluation of qos in the downlink of ofdma wireless cellular networks serving streaming and elastic trac, IEEE Trans. Wireless Commun., 2010. [41] B. Baszczyszyn and M. Karray, Fading eect on the dynamic performance evaluation of OFDMA cellular networks, in Proc. of the 1st International Conference on Communications and Networking (ComNet), 2009.
140
BIBLIOGRAPHY
[42] A. Dembo and O. Zeitouni, Large Deviations Techniques and Applications. Stochastic Modelling and Applied Probability, Springer, 2009. [43] M. Haenggi, A Geometric Interpretation of Fading in Wireless Networks: Theory and Applications, IEEE Transaction on Information Theory, vol. 54, pp. 55005510, Dec. 2008. [44] J. M. Kelif, M. Coupechoux, and P. Godlewski, Spatial outage probability for cellular networks, in GLOBECOM, pp. 44454450, 2007. [45] R. K. Ganti and M. Haenggi, Spatial and temporal correlation of the interference in aloha ad hoc networks, IEEE Communications Letters, vol. 13, pp. 631633, September 2009. [46] X. Lagrange, Cir cumulative distribution in a regular network, tech. rep., ENST Paris, 2000. [47] Comparison of Various Frequency Reuse Patterns for WiMAX Networks with Adaptive Beamforming, 2008. [48] M. Haenggi and R. Ganti, Interference in Large Wireless Networks, Foundations and Trends in Networking, vol. 3, no. 2, pp. 127248, 2008. [49] M. Haenggi, J. G. Andrews, F. Baccelli, O. Dousse, and M. Franceschetti, Stochastic geometry and random graphs for the analysis and design of wireless networks, IEEE Journal of Selected Areas in Communications, vol. 27, pp. 10291046, September 2009. [50] F. Baccelli and B. Blaszczyszyn, Stochastic geometry and wireless networks, volume 2: Applications, Foundations and Trends in Networking, vol. 4, no. 1-2, pp. 1312, 2009. [51] S. Srinivasa and M. Haenggi, Distance Distributions in Finite Uniformly Random Networks: Theory and Applications, IEEE Transactions on Vehicular Technology, vol. 59, pp. 940949, Feb. 2010. [52] S. Foss and S. Zuyev, On a voronoi aggregative process related to a bivariate poisson process, pp. 965981, 1996. [53] P. Robert, Stochastic Networks and Queues, vol. 52 of Applications of Mathematics. Berlin: Springer-Verlag, rst ed., 2003. [54] K. Park and W. Willinger, Self-Similar Network Trac and Performance Evaluation. New York, NY, USA: John Wiley and Sons, Inc., 1st ed., 2000. [55] P. Kuusela, I. Norros, J. Lapuyade-Lahorgue, and M. Naldi, On reliability modeling with stationary on/o processes, Submitted for publication. [56] M. K. Karray, Analytic evaluation of wireless cellular networks performance by a spatial Markov process accounting for their geometry, dynamics and control schemes. PhD thesis, Ecole Nationale Suprieure des Tlcommunications, 2007. [57] S. G. Eick, W. A. Massey, and W. Whitt, The physics of the mt /g/ symbol queue, Oper. Res., vol. 41, pp. 731742, July 1993.
141 [58] S. G. Eick, W. A. Massey, and W. Whitt, mt /g/ queues with sinusoidal arrival rates, Management Science, vol. 39, no. 2, pp. 241252, 1993. [59] W. A. Massey and W. Whitt, Networks of innite-server queues with nonstationary poisson input, Queueing Syst., vol. 13, no. 1-3, pp. 183250, 1993. [60] B. Baszczyszyn and M. Karray, Impact of mean user speed on blocking and cuts of streaming trac in cellular networks, in Proc. of European Wireless Conference, (Prague), 2008. [61] M. Mandjes and P. Mannersalo, Queueing systems fed by many exponential ono sources: an innite-intersection approach, Queueing Syst. Theory Appl., vol. 54, pp. 520, September 2006. [62] A. Brandt and M. Brandt, On the distribution of the number of packets in the uid ow approximation of packet arrival streams, Queueing Syst., vol. 17, no. 1-2, pp. 275315, 1994. [63] H. Kuo, Gaussian Measures in Banach Spaces. Springer-Verlag, 1975. Lecture Notes in Mathematics,
142
BIBLIOGRAPHY

Vu Thesis

Transféré par

Informations du document

Titre original

Copyright

Formats disponibles

Partager ce document

Partager ou intégrer le document

Options de partage

Avez-vous trouvé ce document utile ?

Ce contenu est-il inapproprié ?

Droits d'auteur :

Formats disponibles

Vu Thesis

Transféré par

Droits d'auteur :

Formats disponibles

2012-ENST-00xx

Doctorat ParisTech THSE

Modles spatiaux pour la planication cellulaire

Directeurs de thse: Laurent DECREUSEFOND Philippe MARTINS

Jury Mme/M. Mme/M. Mme/M. Mme/M. Mme/M. Mme/M. Mme/M. Mme/M.

Prnom Prnom Prnom Prnom Prnom Prnom Prnom Prnom

NOM, NOM, NOM, NOM, NOM, NOM, NOM, NOM,

Titre, Titre, Titre, Titre, Titre, Titre, Titre, Titre,

Unit Unit Unit Unit Unit Unit Unit Unit

recherche, recherche, recherche, recherche, recherche, recherche, recherche, recherche,

Ecole Ecole Ecole Ecole Ecole Ecole Ecole Ecole

Fonction Fonction Fonction Fonction Fonction Fonction Fonction Fonction

Spatial models for cellular network planning

Mathematical background and contributions

Dimensioning and coverage models

8 4.4 4.5 4.6 5 On 5.1 5.2 5.3 5.4 5.5

Energy consumption models

Thesis outline and contributions

0 < lim inf xa

Abbreviation LHS RHS PDF CDF CCDF BS

Mathematical background and contributions

Poisson point process

The use of Poisson point process on wireless modelling

2. Poisson point process Realistic modeling

Here we have the correspondence: Analysis f (x) f (0)

Poisson point process

P(w(A1 ) = k1 , ..., w(An ) = kn ) =

2. Poisson point process

Malliavin calculus on Poisson point process

fn (z1 , , zn )( d w(z1 ) d (z1 )) ( d w(zn ) d (zn ))

It is well known that, E

f (z1 , , zn ) d (z1 ) d (zn )

fn (z1 , ..., zn )gn (z1 , ..., zn ) d (z1 )... d (zn )

For fn L2 ( )n and gm L2 ( )m , we dene fn l k gm , 0 l k , to be the function: (yl+1 , , yn , xk+1 , , xm )

fn (y1 , , yn )gm (y1 , , yk , xk+1 , , xm ) d (y1 ) . . . d (yl ). (2.3)

Proposition 1. For fn L2 ( )n and gm L2 ( )m , we have In (fn )Im (gm ) = where hn,m,s =

belongs to L2 ( )n+ms , 0 s 2(m n).

2. Poisson point process

Let D be dened by D : domD L2 (E E, P ), such that F = E [F ] +

nIn1 (fn (, z )).

if and only if the chaos representation of F has the form

2. Poisson point process

Distribution of linear functional of Poisson point process

We call F a linear functional of w if there exists f : E R such that F =

Theorem 8. ([9]) The Laplace functional of w satises: Lw (u) = e

Bn (a1 , a2 , ..., an ) n n! n=1

for i = 1, 2, ..., n. Proof. We apply the theorem 8 to get that: E eF =e =e =e

2. Poisson point process

Campbells formula is written: E [F ] =

E [f (z, m)] d (z ), E f 2 (z, m) d (z )

Existing results on Gaussian approximation, Edgeworth expansion and concentration inequality

We now apply theorem 6 to

2. Poisson point process

ef (z ) f (z ) 1 Thus, P(F > E [F ] + a) exp = exp Minimizing the LHS by =

For > 0, let N =

m(4, 1) m(3, 1)2 + 6 9

Proof. According to the Taylor formula, Dx G(N ) = G(N + f (x)) G(N )

1 2 1 = G (N )f (x) + f (x) G (N ) + f (x)3 2 2 Hence, according to (2.5) and (2.24), E N (Pt G) (N ) = E

r 2 G(3) (rN + (1 r )f (x)) d r. (2.24)

f (x)Dx (Pt G) (N ) d (x) 1 2

(Pt G)(4) (rN + (1 r )f (x))r 2 d r

Hence, e4t |A3 | 6 1 e2t 2 m(4, ) G(3)

2. Poisson point process