Vous êtes sur la page 1sur 70

UMTS Capacity Estimation

Contents
1 UMTS Service Model................................................................................................................................... 1
1.1 Service Classification.......................................................................................................................... 1
1.2 Service Model ..................................................................................................................................... 2
1.2.1 Classification of Area Types.................................................................................................... 2
1.2.2 CS Domain Service Model ...................................................................................................... 3
1.2.3 PS Domain Service Model....................................................................................................... 4
2 UMTS Capacity Estimation........................................................................................................................ 9
2.1 Capacity Estimation Flow................................................................................................................... 9
2.2 Estimation Method of Hybrid Service Capacity................................................................................. 9
2.2.1 Equivalent Erlang Method..................................................................................................... 10
2.2.2 Post Erlang-B Method ........................................................................................................... 11
2.2.3 Campbell Method .................................................................................................................. 12
2.3 Uplink Capacity Estimation.............................................................................................................. 15
2.3.1 Load Analysis for Uplink....................................................................................................... 15
2.3.2 Uplink Capacity and Scale Estimation .................................................................................. 17
2.4 Downlink Capacity Estimation......................................................................................................... 19
2.4.1 Analysis of Downlink Load................................................................................................... 19
2.4.2 Downlink Capacity and Scale Estimation ............................................................................. 22
3 Scale Estimation Example......................................................................................................................... 25
3.1 Assumed Conditions ......................................................................................................................... 25
3.2 Estimation Process............................................................................................................................ 26
3.2.1 Estimation Flow Chart........................................................................................................... 26
3.2.2 Uplink Coverage Estimation.................................................................................................. 26

3.2.3 Uplink Capacity Estimation .................................................................................................. 28


3.2.4 Downlink Capacity Estimation ............................................................................................. 30

ii

1 UMTS Service Model


1.1 Service Classification
In the 3GPP protocol, services running in the UMTS system are grouped into four
classes based on Quality of Service (QoS), Conversational, Streaming, Interactive and
Background. The 3GPP protocol 22.105 gives examples to illustrate typical
applications of these services and their QoS requirements. Table 1.1-1 offers the basic
features and typical cases of these four classes of services.
Table 1.1-1 Service Classification
Service Category
Conversational

Streaming

Interactive

Background

Basic Features

Typical Cases

Keep time relationship between information

Voice

entities in stream, conversational mode (small

conference,

delay and strict delay jitter requirement)

game, Telnet

Keep time relationship between information


entities in stream

Voice

service,

video

interactive

stream

media

download, movie browse,


Video On Demand (VOD)

Request response mode and keep data


integrity

Basic

browse,

mobile

office, information service


and e-commerce

Target has high tolerance to data delay and

Fax service, SMS, MMS,

data integrity shall be kept

FTP, E-mail

For different service cases, the UMTS service bear rate shall be designed according to
their special demands on radio resources. Table 1.1-2 lists the radio bear rates of these
typical services.
Table 1.1-2 Data Service Application Proportion and Bear Rate
Bear Rate (kbps)

Service

Uplink

Downlink

E-mail

64

64

MMS

64

64

Intranet

64

128

E-commerce

64

128

Info Services

64

128
1

Entertainment

64

128

WWW

64

128

FTP

64

128/384

Video streaming

64

384

1.2 Service Model


Service model is the reference for capacity estimation. It reflects the proportion of each
service in hybrid service under various service environments. Based on this proportion,
you can estimate the average traffic or data throughput of a single user. Multiply the
value by the expected number of users in various environments to get the
corresponding total traffic or throughput.

1.2.1 Classification of Area Types


Service model is very important to the UMTS network design because it is the
reference for capacity estimation and determines whether to take future network
service demands into account during planning. On the other hand, service model is
hard to predict. Service model is closely associated with the behavior habits of different
users using different services and users habits of using services are closely associated
with many factors in different areas, such as economy and culture. Therefore, a service
model is inapplicable for the application requirements of different environments.
According to service type distribution, service development policy and user dynamic
distribution as well as consumption behavior features in an area, service distribution
areas are categorized into six classes, downtown area, urban area, suburb area, rural
area, main line of communication/scenic spot and indoor coverage. Table 1.2-1 gives
service distribution features and user density of different areas.
Table 1.2-1 Service Distribution Features and User Density of Different Areas
Service
Area

Distribution
Feature

Downtown area

Site
Classification

Traffic-intensive

Central business

High service rate

district*

requirement

Irregular

Key area of data

building-intensiv

service

e area

User Density
2

(user/km )
>12000

Population Density
2

(user/km )
>50000

>8000
>30000

development

Dense

building

>1000

complex area
High traffic; medium service rate;

Urban area

common data service demands


Low traffic; low-speed or no data

Suburb area

service
Sparse traffic; with the purpose of
solving coverage

Rural area

>1000
<1000

<1000

>10000
>3000
>100

<100

No guarantee for data service QoS


Main

line

of

communication/
Scenic spot

Low traffic; scenic spot with seasonal


features

Note: *Such regions as Middle Ring of HongKong and Lujiazui Financial & Trade Zone in Shanghai

Considering voice service remains important at the early stage of 3G construction, the
following sections gives detailed recommendation of busy hour traffic for the above six
areas respectively on voice service and video phone service; with respect to PS domain
data service, busy hour traffic for only the former four areas are provided.

1.2.2 CS Domain Service Model


Figure 1.2-1 shows the voice service call model.

Call duration

Call setup

Call release

Figure 1.2-1 Call Model of Voice Service

Primary parameters of the voice service model are Busy Hour Call Attempt (BHCA)
and call duration. With these two parameters, you can calculate the busy hour traffic.
Busy Hour Traffic = BHCA x Call Duration /3600
Tables 1.2-2 and 1.2-3 respectively offer the recommended values of voice service and
video phone service in different areas.
Table 1.2-2 Voice Service Model

Area

BHCA

Central business

Call Duration (S)

Traffic (Erl/BH)

2.7

60

0.045

1.8

60

0.03

1.2

60

0.02

Urban area

1.2

60

0.02

Suburb area

1.018

60

0.018

Rural area

0.96

60

0.016

0.9

60

0.015

district
Downtown
area

Irregular
building-intensiv
e area
Dense

building

complex area

Main

line

of

communication/scenic spot

Table 1.2-3 Video Phone Service Model


Area

BHCA

Central business

Call Duration (S)

Traffic (mErl/BH)

0.135

120

4.5

0.09

120

0.06

120

Urban area

0.06

120

Suburb area

0.0509

120

1.8

Rural area

0.048

120

1.6

0.045

120

1.5

district
Downtown
area

Irregular
building-intensi
ve area
Dense building
complex area

Main

line

communication/scenic spot

of

1.2.3 PS Domain Service Model


The data service call model widely differs from the voice service call model. Data call
has the following features:
Conversion between Dormant state and Active state;
Each session of a user can consist of several packet calls and different data service
types and user types have differentiated features;
Data is transmitted in data burst mode;
Resources occupied by packet call vary with data burst transmission.

Figure 1.2-2 shows the data call process:


A data service process of a user

A data service process of a user

Data call(session)-WWW
Click web
page

Click next page

Packet Call

Packet Call
Web page
download

Web page download

Active

Data call(session)-WWW
Click next
page

Dormant

Send/Receive Email

Packet Call

Active Dormant

Web page download

Data Burst

Web page download

Active

Active

Next web page download


Packet Call

Packet Call

Data Burst

Packet Call

Web page download

Data Burst

Data Burst

Call setup

Call release

Active

Dormant

Figure 1.2-2 Data Service Call Process

The data service here is described in ETSI model. Its primary parameters are Busy
Hour Session Attempt (BHSA), calls per session, packets per call and mean packet size.
With these parameters, you can work out the busy hour service throughput and
equivalent Erl. Table 1.2-4 gives the calculation method of data service throughput.
Table 1.2-4 Data Service Throughput Calculation
Parameter

Symbol

BHSA of data service

Application proportion

Calls per session

Packets per call

Mean packet size (Byte)

Service bear rate

Service throughput (kbits/BH)

G = a*b*c*d*e*8/1000

Erl

h = g/3600/f

Based on the national CDMA user habit analysis statistics in conjunction with the

international UMTS data service features, parameters of data service ETSI model in
downtown area are given in Table 1.2-5:
Table 1.2-5 Parameters of Data Service ETSI Model in Downtown Area

Service

BHSA

Call per

Packet

Session

in a Call

UL/DL

UL/DL

Mean Packet

Throughput

Size (Byte)

UL/DL(kbits)

E-mail

0.3

2/2

15/15

480

34.56/34.56

MMS

0.05

2/2

15/15

480

5.76/5.76

Intranet

0.15

5/5

4/27

480

11.56/77.76

0.05

2/2

10/26

480

3.84/9.98

0.08

2/2

5/33

480

6.14/40.69

0.02

5/5

4/27

480

1.54/10.37

WWW

0.2

5/5

2/15

480

7.68/57.60

FTP

0.15

1/1

8/74

480

4.61/42.62

E-com
merce
Info
Services
Entertai
nment

Because all services will finally come down to the bear rate, Table 1.2-6 provides a
recommended data service model at the early stage of 3G construction based on bear
rate. Where, 384 service is applicable only for downtown and urban areas due to its
great impact on network coverage.
Table 1.2-6 Data Service Model
Bear

Busy Hour Traffic (kbits)

Rate

Downtown

(kbps)

Area

Urban Area

Suburb Area

Uplink/Downli
Rural Area

nk Proportion

64/64

80.64

63.04

38.8

15.76

1:1

64/128

161.88

140.3

87.35

34.94

1:7

64/384

112.51

86.8

54.25

21.7

1:10

Note: The data in this table is intended for Class 4 area, which relatively drops behind
Class 1, 2 and 3 areas so that you can multiply the data by 30, 20 and 10 respectively
for these areas. Overseas developed areas are taken as Class 1 areas.

From analysis, 31 provinces and cities in China mainland can be categorized into four
6

regions. The telecommunication development in provinces and cities of the same


region has many similarities so that they are taken into account comprehensively.
These four classes of regions are as follows:
Class 1: Guangdong, Shanghai, Beijing and Zhejiang;
Class 2: Tianjin, Fujian, Shandong, Liaoning, Sichuan, Chongqing and Jiangsu;
Class 3: Heilongjiang, Jilin, Hunan, Inner Mongolia, Hubei, Henan and Hainan;
Class 4: Qinghai, Hebei, Tibet, Shanxi, Anhui, Guangxi, Jiangxi, Shaanxi, Yunnan,
Gansu, Ningxia, Xinjiang and Guizhou.

2 UMTS Capacity Estimation


2.1 Capacity Estimation Flow
The capacity estimation is another important part of the scale estimation. The purpose
of capacity estimation is to estimate the approximate BS number needed by the
capacity according to the service model and service traffic demand of the network
planning. Similar with the link budget, the capacity estimation should be performed
from the uplink and downlink. For the UMTS system capacity, the interference is
limited in the uplink direction and the BS power is limited in the downlink direction. In
the 2 G CDMA network, the voice service is the main application service with
symmetrical uplink and downlink traffic, the capacity is limited in the uplink direction,
so the uplink capacity calculation is focused on in capacity estimation. However, in the
UMTS network, the data service proportion is obviously increased and the network
uplink and downlink traffic becomes asymmetric generally, and even the downlink
capacity may be limited. Therefore, the UMTS capacity estimation should be
performed from the uplink and downlink respectively. The following steps are involved
in capacity estimation:
1

Hybrid service intensity analysis. The UMTS system can provide multiple
services. The hybrid service intensity analysis makes the system capacity
consumed by various services equivalent to that consumed by a single service.

Uplink capacity estimation. Estimate the BS number that meets the service
demand based on the hybrid service intensity analysis.

Downlink capacity estimation. It is a verification process. The BS transmission


power formula is used to calculate the channel number that can be provided by
the current BS scale so as to verify whether this channel number can meet the
capacity requirement, and if it cannot, stations need be added.

2.2 Estimation Method of Hybrid Service Capacity


There are multiple services in the UMTS network, their service rates and required
Eb/No are diversified, the effects on the system load and consumed BS resources are
different, so the estimation for the cell capacity cannot adopt the method for estimating

the cell capacity in a pure voice network. An idea of hybrid service capacity estimation
is to make equivalent among various services to make the system capacity consumed
by various services equivalent to that consumed by a single service. The Equivalent
Erlang, Post Erlang-B and Campbell methods in the hybrid service estimation are
introduced respectively as follows.

2.2.1 Equivalent Erlang Method


The fundamental principle of the Equivalent Erlang method is to make a service
equivalent to another service, calculate the total traffic (erl) of the equivalent services
and count the channel number needed by this traffic. We will give an example to
explain it as below.
Suppose services A and B are provided in the network, where,
service A: each connection occupies one channel and the total is 12 erl;
service B: each connection occupies 3 channels and the total is 6 erl.
If 1 erl service B is equivalent to 3 erl service A, the total traffic in the network will be
12+6*3=30 erl (service A). After querying Table erl-B, we know that altogether 39
channels are needed under 2% blocking rate.
If 3 erl service A is equivalent to 1 erl service B, the total traffic in the network will be
12/3+6=10 erl (service B). After querying Table erl-B, we know that altogether 17
service B channels (equivalent to 17*3=51 service A channels) are needed under 2%
blocking rate.
Upon the above analysis, we know that calculation result through the Equivalent
Erlang method is related to the equivalent mode adopted. The result through the former
equivalent mode is too small (39 channels) which is too optimistic, while the result
through the latter mode is too large (51 channels), which is too pessimistic, as shown in
the following figure:

10

Capacities meeting the


same GOS are different
Low speed
service
equivalent

2 Erl Low
speed service

1 Erl High
speed service
High speed service
equivalent

The calculation
result is related
to the
equivalent
mode

2.2.2 Post Erlang-B Method


The fundamental principle of the Post Erlang-B method is to calculate the channel
number required by each service capacity respectively and add channels in an
equivalent manner to obtain the channel number required by the hybrid service capacity.
We will give an example to explain it as below.
Suppose services A and B are provided in the network, where,
service A: each connection occupies one channel and the total is 12 erl;
service B: each connection occupies 3 channels and the total is 6 erl.
After querying Table erl-B, we know that altogether 19 channels are needed to meet
service A traffic (12 erl) under 2% blocking rate.
After querying Table erl-B, we know that altogether 12 service B channels (equivalent
to 12*3=36 service A channels) are needed to meet service B traffic (6 erl) under 2%
blocking rate.
The two services need 19+36=55 channels totally.
Calculate the network capacity in a special case based on the Post Erlang-B method:
Suppose services A and B are the same kind, where,
service A: each connection occupies one channel and the total is 12 erl;
service B: each connection occupies 1 channels and the total is 6 erl.
After querying Table erl-B, we know that altogether 19 channels are needed to meet

11

service A traffic (12 erl) under 2% blocking rate.


After querying Table erl-B, we know that altogether 12 channels are needed to meet the
service B traffic (6 erl) under 2% blocking rate.
Services A and B need 19+12=31 channels totally.
Because services A and B are the same kind, the total traffic is 12+6=18 erl. According
to the currently known method of capacity calculation in single service, after querying
Table erl-B, we know that 26 channels are needed to meet the traffic demand under 2%
blocking rate. This result is correct obviously.
Upon above analysis, we can see that the calculation result through the Post Erlang
method is too pessimistic (31>26). The reason is that the BS channels are shared
among services, however, the Post Erlang method factitiously separates the channels
used by the services, and thus, the BS channel resource utilization ratio is reduced, as
shown in the following figure:
Capacities meeting the same
GOS are different

1 ERL service A

1 ERL service B

1 ERL service A and


1 ERL service B

The
calculation
result is too
pessimistic

2.2.3 Campbell Method


The fundamental principle of the Campbell method is to make all services equivalent to
a virtual service based on certain rules, calculate the total traffic (erl) of this virtual
service, count the virtual channel number needed by this traffic, and convert the
number into the actual channel number that meets the network capacity.
The equivalent principle of the Campbell model:

12

erl a
v
c
erl a

i i

i i

OfferedTraffic
Capacity

(Ci ai )
c

Where, c indicates capacity factor.


v indicates hybrid service variance.
indicates hybrid service mean.

ai
Ci

indicates the equivalent intensity of service i.


indicates the channel number needed by service i.

OfferedTraffic indicates traffic of the virtual service.


Capacity indicates the virtual channel number needed by the virtual traffic.
We will give an example to explain it as below.
Suppose services A and B are provided in the network, where,
service A: each connection occupies one channel and the total is 12 erl;
service B: each connection occupies 3 channels and the total is 6 erl.
Equivalent intensity of service A a1=1 and that of service B a2=3.

erl a

The hybrid service mean is

The capacity factor is

The virtual traffic is

12163 30

The hybrid service variance is

i i

erl a

i i

12 16 3 66

66
30

2.2

OfferedTraffic

30

13.63
c 2.2

After querying Table erl-B, we know that altogether 21 virtual channels are needed to
meet the virtual traffic under 2% blocking rate.

13

According to formula (), under 2% blocking rate, the channel number needed by each
service is shown as follows:
Service A:
Service B:

C1 (212.2) 1 47
C1 (212.2) 3 49

From the above analysis, compared with results of the Equivalent Erlang and Post
Erlang-B methods, the result of the Campbell method is more credible, so it is a more
reasonable estimation method for hybrid service capacity at present. According to the
Campbell method, under the same requirement of the service level GOS, diversified
channel resources are needed by different services, or, under the same channel
resources, different services obtain diversified service levels. From this point of view,
the Campbell method is more reasonable. However, the Campbell method makes all
services uniformly equivalent as the circuit domain services and uses the Erlang-B
model for analysis and calculation. In fact, the features of the packet domain services
are completely different from those of the circuit domain services, and in addition, the
Erlang-B establishment conditions are not satisfied, so this equivalent method has
defects itself. A further research is needed for better hybrid service establishment
model and capacity analysis method.
In the Campbell method, the service equivalent intensity a can be calculated based on
channel number consumed by each kind of service or based on the interference
introduced from the air interface by each kind of service, shown as follows:

Eb
bit rate for service

Relative amplitude

bit rate for amplitude1

Eb

for service

N0
N0

for amplitude1

If the reference service is the voice service, with its activity at the physical layer
considered, the above formula can be modified to:

Relative amplitude

Eb
bit rate for service
Eb
bit rate for voice

N0

N0

for service

for voice v for voice

14

2.3 Uplink Capacity Estimation


2.3.1 Load Analysis for Uplink
In the UMTS system, all users adopt the same carrier and each signal becomes a noise
(interference) for others upon coding. Therefore, each signal is contained in the
bandwidth interference background generated by other user. To access a call, the
mobile station power must be large enough to overcome other mobile stations in the
bandwidth, that is, the receive signal in the BS must reach Eb/No (energy per user bit
to noise spectral density) required by the service demodulation.

(Eb / No) j

User J's handling gain

User J's signal


Total receive power (without its own signal)

The above formula can be written into:

(Eb / No) j

Pj

v j R j Itotal Pj

Where, W indicates the chip rate, 3.84 Mchip/s.


vj indicates user js activation factor.
Rj is user js bit rate.
Pj indicates receive power for signals from user j.
Itotal indicates total broadband receive power with the thermal noise power included of
the BS.
From the above formula, we know that the receive power at the BS receive end should
meet the following formula so that the user signal can meet the demodulation
requirement:

Pj

1
Itotal
W
1
(Eb / No) j R jv j

Define a connection load factor Lj:

Lj

1
W
1
(Eb / No) j R j v j

Lj indicates the ratio of user signal power to the total BS receive power, so a single

15

user signal power Pj is represented to

Pj L j Itatal

The total receive power of all N users from one cell is:

P L
N

j1

j Itatal

j1

Generally, the total receive power at the BS receive end consists of in-cell user
interference power, out-cell user interference power and BS thermal noise, that is:

Itatal Pin Pother PN


Where, Pin indicates the total interference power of in-cell users.
Pother indicates the total interference power of out-cell users.
PN indicates the BS thermal noise power.
Because the out-cell mobile station interference power is not controlled by the local
cell BS, the interference is hard to determine. Generally, define the ratio of the
interference from other cell to that of the local cell as the neighbor cells interference
factor i:
i

Other cell interference


Local cell interference

i indicates the ratio of other cell interference to the local cell interference at the BS
receive end of the local cell. Generally, the neighbor cell interference factor of the
macro cell that adopts omni antenna is 0.55 and that of the macro cell that adopts
three-sector antenna is 0.65.
Therefore, the total user receive power of the BS is

Pin Pother (1i)

j Itatal

j1

Define the noise lifting as the ratio of total broadband receive power to the noise power
of the BS, that is,

Itotal
NR P I
N

Itatal
P P

tatal

in

other

L
N

1(1i)

j1

UL

Define the uplink load factor

as
16

UL (1i)

1
W
j1 1
(Eb / No) j R jv j

(1i)

j1

F
o

r
UL indicates the ratio of the user signal power at the BS receive end to the total
m
receive power of the broadband.
u
Then, the noise lifting can be represented to
l

NR

1
1UL or

NR(dB) 10LOG10(1UL )

(
This equation reflects the thermal noise lifting caused by user interference at the BS
1
receive end. 3 dB noise lifting corresponds to 50% load factors and 6 dB noise lifting
)
corresponds to 75% load factors. Generally, the network planning supposes that the
uplink load factor is 50%, in a single service, the channel number provided by each cell
can be calculated through formula (1), and then, the total BS number required by the
uplink capacity demand can be counted further. For the capacity estimation for hybrid
service, the Campbell algorithm should be combined to make the system resources
consumed by various services equivalent to those consumed by a single service. Then,
the channel number provided by each cell can be calculated through formula (1), and
the BS number required by the hybrid service capacity demand can be counted further.
The next section details the capacity estimation flow of the hybrid service.
The uplink noise lifting NR corresponds to the interference margin in the uplink
budget, that is, the coverage is related to the capacity. In planning, the network load
factor should be determined to get the noise lifting corresponding to this load. Then,
the BS radius meeting the uplink capacity requirement can be calculated further
through the link budget.

2.3.2 Uplink Capacity and Scale Estimation


The previous section describes the load factor of uplink, based on which, this section
describes how to estimate the BS quantity satisfying the composite traffic requirements
for uplink. Figure 2.3-1 shows the flow of estimating uplink capacity.

17

Calculate the quantity

Calculate equivalent

of equivalent voice channels

intensity of services

in a cell

Calculate the variance,

The quantity of virtual

average value and capacity

channels in the sell

factor of the mixed service

Virtual traffic A of the

Virtual traffic B of the cell

system

A/B

Number
of cells

Figure 2.3-1 Flow Chart of Estimating Uplink Capacity

Calculate the virtual composite traffic of the system.


Because various services have different effects on system load, such an effect
can be equivalent to the effect of multiple voice channels on system load. The
calculation formula is as follows:
amplitude service= (Rservice x Eb/Noservice x vservice)/ (Rvoice x Eb/Novoice
x voice)
Where, R represents service rate.
Eb/No represents quality factor of the service.
v represents the activation factor of the service at the physical layer
According to the Campell theory, the virtual composite traffic of the system can
be calculated.

Calculate the quantity N of equivalent voice channels provided by a cell.


Suppose that system capacity load is represented by , the uplink capacity

18

formula is as follows
N

(1 f ) *

1
W 1
1
1 * *
R v j Eb
No

Where, represents load factor, f represents interference factor from an adjacent


cell, v represents activation factor and N represents the quantity of channels.
According to the above formula, the quantity N of equivalent voice channels
provided by a cell can be evaluated.
3

Calculate the quantity of virtual channels in every cell


Bases on the quantity N of equivalent voice channels evaluated in step 2 and the
following formula
voice channel=virtual channel*C
Where, voice channel is the quantity N of equivalent voice channels
The quantity of virtual channels in every cell can be evaluated.

Look up Table Erl B according to the quantity of virtual channels evaluated in


step 3, and get the quantity of virtual traffic in every cell.

Calculate the quantity of cells


According to the virtual composite traffic of the system evaluated in step 2 and
virtual traffic of every cell evaluated in step 4, calculate the quantity of required
cells:
the number of cells=composite traffic/the virtual Erlang number every cell
The quantity of BSs required in three sectors is calculated as follows: the
number of cells/3.

2.4 Downlink Capacity Estimation


2.4.1 Analysis of Downlink Load
On the downlink, BS power is shared by all users in a cell. When no power in total BS
power can be allocated to a new user, air interface capacity reaches its limit. That is to
say, when a BS transmits the total power used for normal running of all users exceeds

19

the rated power of the BS, downlink capacity reaches power limit. Therefore, downlink
capacity is limited by the total transmitting power of the BS.
Similar to the analysis method of uplink capacity, analysis of downlink capacity starts
from the Eb/No value required by signal demodulation. To correctly demodulate useful
signals on the downlink, the mobile station must overcome interference from the
following three aspects: interference caused by nonorthogonality of the channel in a
cell, interference of signals from the outside of the cell and thermal noise from the
mobile station. That is,

Itatal (1)P Pother PN


Where, P represents total BS transmitting power.
Pother represents total interference power of signals from the outside of the cell.
PN represents thermal noise power from the mobile station.

represents quadrature factor of the downlink.


On the downlink, quadrature factor is a very important parameter. Users on the
downlink of UMTS are differentiated based on the orthogonal code. In the case there is
no multipath propagation, orthogonality keeps unchanged when the mobile station
receives signals from the BS. However, in the actual process of signal propagation,
multipath delay is unavoidable, therefore orthogonality between channels is damaged,
thus causing interference. Quadrature factor 1 corresponds to the user fully orthogonal.
Generally, quadrature factor in the multipath channel is between 0.4 and 0.9.
By referring to the derivation means of uplink load factor, denote the downlink load
factor

DL

as follows:

(Eb / No)

DL

j1

W/Rj

[(1j ) i j ]

Where, W represents chip rate at 3.84 Mchip/s.

vj

represents activation factor of the user j.

Rj
j

represents bit rate of the user j.


represents channel quadrature factor from the user j.

i j represents the ratio of BS power received by the user j from other cell to that from

20

this cell.
Because mobile stations are distributed randomly in a cell,

and i j are related to

the location of users. For the average value of cell load factors, adopt its similar
average value in the whole cell, that is:
N

DL

j1

(Eb / No) j
W/Rj

[(1) i]

Where, represents the average quadrature factor in a cell. Generally, it is 60% for
the multipath channel and 90% for the non-multipath channel. i represents the
average ratio of the BS power received by the user from other cell to that from this cell.
Generally, it is 55% for the omni antenna macro cell and 65% for the three-sector
antenna macro cell.
During the analysis of downlink capacity, estimation of BS transmitting power is the
most important. The estimated BS transmitting power is average power not peak power
at the cell boundary, because the transmitting power distributed by the BS for each user
is determined by the average loss from the BS to the mobile station and the sensitivity
of the mobile station. On the actual network, users are distributed randomly in a cell,
not at the cell boundary, therefore, the average path loss value, not the maximum path
loss value estimated for the link, should be adopted when BS transmitting power is
calculated. In a macro cell, the difference between the maximum path loss and the
average path loss is usually 6 dB.
The total BS transmitting power can be expressed by the following formula:

(Eb / No)

NrfW L

j1

BS _ TxP
Where,

Nrf

W/Rj

1 DL

F
o

represents the noise power spectrum density on the front of the mobiler

station receiver, and it can be calculated by the following formula:

m
u

Nrf KT NF 174.0dBm NF (sup poseT 290K) Where, NF

l
represents the noise coefficient of the mobile station receiver with the typical value of 5
a
dB to 9 dB.
L represents the average path loss, which is evaluated by subtracting 6 dBm from the(

maximum path loss.

)
21

vj represents activation factor of the user j.


Rj represents bit rate of the user j.
In the case of a single service, evaluate the channel quantity provided by every cell
under the maximum allowed transmitting power according to the formula (2) and
further evaluate the total number of BSs satisfying downlink capacity requirements.
In fact, the analysis of uplink and downlink link performances is a hard process.
Because the performance of downlink depends on many basic elements very much, its
analysis cannot be streamlined like the analysis of uplink. The Eb/No value range of
downlink is a parameter changing greatly with moving speed and multipath condition.
In addition, the mobile station receiver does not use antenna diversity. The reason why
the required Eb/No value changes with the mobile station is that at least two paths
cannot be ensured unless it is clearly known that the mobile station is in soft handoff or
softer handoff statuses. Such a change, randomicity of mobile station location and
interference level from the surrounding cell make the analysis of downlink
performance complicated. In designing, a very conservative conclusion can be gotten
in the case the worst condition is considered. Generally, estimate capacity after
analyzing the channel quantity required by uplink capacity, and observe whether the
downlink can support the mobile station to work in the designated coverage area and
its channel quantity reaches the channel quantity generated by the uplink.

2.4.2 Downlink Capacity and Scale Estimation


Downlink estimation is a verification process. The process of downlink capacity and
scale estimation is as follows: First calculate the quantity of equivalent voice channels
to be provided by this cell in the current service model, and then calculate the quantity
of equivalent voice channels availably provided by the cell according to the downlink
power calculation formula, and subsequently compare these two results. If the quantity
to be provided by the cell is less than that availably provided by the cell, it indicates
that downlink power is enough and the current scale satisfies system capacity
requirements. If the former is larger than the latter, it indicates that downlink capacity
is limited. To make downlink power enough, add some BSs.
1

Calculate the quantity of equivalent voice channels to be provided by every cell.


Under the precondition of known reverse capacity and scale, you can evaluate
the traffic of various services in every cell under such a scale. Then, according

to the equivalence of voice channels, you can evaluate the quantity of equivalent
22

voice channels to be provided by every cell. This quantity can be calculated by


following several steps below
1)

Calculate the average traffic of various services in every cell according to the BS
quantity of uplink and total traffic of downlink.
Average

traffic

of

various

services

in

cell

TotalDownlinkTraffic
UplinkBaseStationQuantity 3
Where, the BS quantity is the larger value between estimated uplink coverage
and estimated capacity result.
2)

According to the Campell theory, calculate the virtual Erlang traffic in every cell.

The calculation method in this step is the same as that of uplink.


3)

Look up Table Erl B according to the virtual Erlang traffic in every cell
evaluated in step 2, and calculate the quantity of virtual channels in every cell.

4)

According to the quantity of virtual channels evaluated in step 3 and the


following formula

(Ci ai )

Capacity

you can evaluate the quantity of equivalent voice channels to be provided by


every cell.
2

Calculate the quantity of equivalent voice channels availably provided by the


cell.
According to the forward power formula
N

PN * L *

j1

vj*

j1

(Eb / No) j
*
W / Rj

(Eb / No) j
W / Rj

[(1j ) j ]

Where, PN represents the noise power spectrum density on the front of the
mobile station receiver, and it can be calculated by the following formula:

PN KT NF 174.0dBm NF (sup poseT 290K) ,


NF represents the noise coefficient of the mobile station receiver with the typical

23

value of 5 dB to 9 dB.

L represents the average path loss, which is evaluated by subtracting 6 dBm


from the maximum path loss. j represents the average quadrature factor.
Generally, it is 0.6 for the multipath channel and 0.9 for the non-multipath
channel.

j represents interference factor from an adjacent cell. Generally, it is 0.55 for


the omni antenna macro cell and 0.65 for the three-sector antenna macro cell.
The quantity of equivalent voice channels availably provided by the cell can be
calculated.
3

Compare the above two results. If the quantity to be provided by the cell is less
than that availably provided by the cell, it indicates that downlink power is
enough and the current scale satisfies system capacity requirements. If the
former is larger than the latter, it indicates that downlink capacity is limited. To
make downlink power enough, add some BSs.

24

3 Scale Estimation Example


3.1 Assumed Conditions
Suppose that the service model data of this area is as follows:
Channel environment: TU 3 km/h
System design load: 50%
Voice service blocking rate: 0.02%
Interference factor from the adjacent cell: 0.65
Quadrature factor: 0.6
Area of the city zone: 40.8 square kilometers
Services in this area are planned as follows:
Uplink:
Voice

CS64

PS64/64

PS64/128

PS64/384

Data rate(k)

12.2

64

64

64

64

Activity factor

0.67

Eb/No

4.7

2.87

1.6

1.6

1.6

Forecast traffic

3000

400

100

Downlink:
Voice

CS64

PS64/64

PS64/128

PS64/384

Datarate(k)

12.2

64

64

128

384

Activity factor

0.58

Eb/No

7.7

7.7

7.4

6.4

Forecast traffic

3000

400

100

40

20

25

3.2 Estimation Process


3.2.1 Estimation Flow Chart
Input:system load requirment and
coverage requirement

Uplink coverage
estimation

Downlink coverage
estimation

Quantity of BSs
satisfying uplink
coverage

Uplink capacity
estimation

Quantity of BSs
satisfying downlink
coverage

Quantity of BSs
satisfying uplink
capacity

Compare the results


and evaluate the
larger one

Based on traffic type

Based on power

Quantity A of
channels to be
provided by every cell
on the downlink

Quantity B of
channels availably
provided by every
cell on the downlink

No
A<B

Yse
End

Figure 3.2-1 Estimation Flow Chart

3.2.2 Uplink Coverage Estimation


1

Evaluate the maximum allowed path loss through link budget

26

The calculation formula of uplink budget is as follows:


The maximum allowed space path loss mobile station transmitting power
(dBm) mobile station antenna gain (dB) human body loss (dB) BS feeder
loss (dB) BS receiving antenna gain (dBi) soft handoff gain (dB) building
or car body penetration loss (dB) slow fading margin (dB) power control
margin (dB) interference margin (dB) BS receiving sensitivity (dBm)
Voice

CS64

PS64

PS64/128

PS64/384

21

21

21

21

21

Antenna gain (dBi)

Human body loss (dB)

Effective transmitting power

19

21

21

21

21

-174

-174

-174

-174

-174

-108

-108

-108

-108

-108

2.2

2.2

2.2

2.2

2.2

Receiver noise (dBm)

-105

-105

-105

-105

-105

Interference margin (dB)

Bit rate (kbit)

12.2

64

64

64

64

Processing gain (dB)

24.98

17.78

17.78

17.78

17.78

Receiving Eb/No (dB)

4.2

2.87

1.6

1.6

1.6

Receiver sensitivity

-124

-118

-119

-119

-119

Antenna gain (dBi)

17

17

17

17

17

Line loss

Power control margin

Soft handoff gain

Shadow fading margin

10.3

10.3

10.3

10.3

10.3

Penetration loss

20

20

20

20

20

121.4

122.7

122.74

122.74

Maximum
Transmitting
end

transmitting

power (dBm)

Thermal

noise

power

spectrum density (dBm/HZ)


Thermal noise power (dBm)
Receiver noise coefficient
(dB)
Receiving
end

Others

Maximum allowed path loss

125.34

Calculate the cell coverage radius according to a specific propagation model


Here, we adopt a universal propagation model of Aircom to calculate:
Path loss = k1 + k2log(d) + k3Hms + k4log(Hms) + k5log(Heff) +
k6log(Heff)log(d) + k7(diffraction loss) + clutter loss
Use parameters in the following table

27

k1

152.4

k2

44.6

k5

-13.82

k6

-6.55

Heff

30

K1 and K2 parameters have greater effect on the budget result. While, K3 and K4 have
less effect, so their values are 0.
Obtain the BS coverage radius after adopting the maximum path loss:

Radius (Km)

Voice

CS64

PS64

0.65

0.5

0.54

PS64/128
0.54

PS64/384
0.54

Calculate the number of BSs required by uplink


From the result in the previous step, we see that the uplink coverage is limited
by the CS64kps service, so the BS radius satisfying successive coverage of
CS64kps is adopted when the number of BSs is calculated.
If the coverage area S of the three-sector BS
0.488 Km

9
8

3R 1.95 0.5

The number of BSs satisfying uplink coverage requirement is 40.8/0.488 = 84


For downlink budget, because all users in the cell share BS power simultaneously, the
cell radius on the downlink is not only related to the number of users in the cell, but
also related to user location and services used by users. The balance between the uplink
and downlink should be calculated iteratively with the planning software. First predict
coverage area for the uplink, and then allocate power for the downlink. If the total
power does not exceed the maximum transmitting power of the BS, links are balanced.
If the total power required by the downlink exceeds the maximum transmitting power
of the BS, coverage area should be reduced and power should be re-allocated to the
downlink until the total power is less than or equal to the maximum transmitting power.

3.2.3 Uplink Capacity Estimation


1

Calculate the virtual composite traffic of the system.

1)

Equivalent service intensity of each service


According to the formula

28

Relativeamplitude

Eb
bit rate for service
Eb
bit rate for amplitude1

N0
N0

for service
for amplitude1

obtain
voice: 1
CS64: 64 x 1 x 100.287/12.2 x 0.67 x 100.42 = 5.76
PS64/64: 64 x 1 x 100.16/12.2 x 0.67 x 100.42 = 4.3
PS64/128: 64 x 1 x 100.16/12.2 x 0.67 x 100.42 = 4.3
PS64/384: 64 x 1 x 100.16/12.2 x 0.67 x 100.42 = 4.3
2)

Calculate the mean of composite traffic

mean erliai 300014005.67 1004.354.324.3 5766.1


i

3)

Calculate the variance of composite traffic


variance erliai 300014005.67 1004.3 54.3 24.3 18271.7
2

4)

Calculate the capacity factor


capacity factor variance/mean 3.17

5)

Calculate the virtual composite traffic of the system


composite traffic mean/capacity factor 5766.1/3.17 1818.96 (Erl)

Calculate the quantity N of equivalent voice channels availably provided by the


cell
According to the uplink load formula
N

(1 f ) *

1
1
1
*
R v j Eb
No

1 W *
j

Where, 50% and f 0.65


get the quantity of equivalent voice channels N 54
3

Calculate the quantity of virtual channels in every cell

29

According to

Capacity

(Ci ai )
c

get the quantity of virtual channels in the cell (54 1)/3.17 16


4

Look up Table Erl B according to the quantity of virtual channels evaluated in


step 3, and get the quantity of virtual traffic in every cell, that is 9.83 Erl.

Calculate the number of cells


Number of cells Virtual traffic of the system/virtual traffic of every cell
1818.96/9.83 186
The number of required three-sector BSs 186/3 62
After the above calculation, we know that 84 stations are required for uplink
coverage. The evaluated number of stations is less than 84, so it meets both
coverage and capacity requirements.

3.2.4 Downlink Capacity Estimation


Downlink capacity estimation is a verification process. With the downlink power
formula, verify whether the number of BSs evaluated from uplink coverage and
capacity budget meets the power requirement. Add BSs until downlink power meets
the requirement.
1

Calculate the quantity of equivalent voice channels to be provided by every cell.

1)

Calculate the average traffic of various services in each cell according to the BS
quantity of uplink and total traffic.
Average traffic of various services in every cell is:
Voice: 3000/84/3 = 11.9 Erl
CS64: 400/84 = 1.59 Erl
PS64/64: 100/84 = 0.4 Erl
PS64/128: 35/84 = 0.14 Erl
PS64/384: 20/84 = 0.079 Erl

2)

Calculate the virtual Erlang traffic in every cell.

30

Equivalent service intensity of each service on the downlink


Voice: 1
CS64: 64 x 1 x 10

0.77

PS64/64: 64 x 1 x 10

0.77

/12.2 x 0.67 x 10

= 7.8

0.74

0.77

PS64/128: 144 x 1 x 10

/12.2 x 0.67 x 10

0.64

= 7.3

0.77

/12.2 x 0.67 x 10

0.8

0.77

PS64/384: 144 x 1 x 10 /12.2 x 0.67 x 10

= 13.1

= 50

The mean of composite traffic is


Mean = 11.9 1 1.59 7.8 + 0.4 7.3 + 0.14

13.1 + 0.079 50

= 33.04
The variance of composite traffic is
Variance = 11.91 1.59 7.82 + 0.4 7.32 + 0.14 13.12 + 0.079
502 = 355.19
Capacity factor = variance/mean = 355.19/33.04 = 10.75
Virtual traffic of the cell
composite traffic = mean/capacity factor = 33.04/10.75 = 3.07 (Erl)
3)

Check Table Erl B and obtain that the quantity of virtual channels required by
every cell is 7

4)

Calculate the quantity of equivalent voice channels required by each cell.


According to the formula

Capacity

(Ci ai )
c

evaluate the quantity of equivalent voice channels is: 7 10.75 1 76.


2

Calculate the quantity of equivalent voice channels actually provided by every


cell.
According to the downlink power formula

31

PN * L *

j1

vj*

j1

(Eb / No) j
vj*
W / Rj

(Eb / No) j
W / Rj

[(1j ) j ]

Where, P represents the maximum service transmitting power, which is 13 W.


PN represents the noise power spectrum density on the front of the mobile station
receiver, and its value is -169 dBm.

L represents average path loss, which is evaluated by subtracting 6 dBm from


the maximum path loss.

represents average quadrature factor, which is 0.6 for the multipath

channel.

represents interference factor from an adjacent cell. It is 0.65 for the

three-sector antenna macro cell.


Obtain that the quantity of equivalent voice channels actually provided by every
cell is 71.
3

Comparison
Through downlink budget, the quantity of channels required by every cell is 76
when there are 84 BSs in a network. However, according to the power formula,
the quantity of channels actually provided by every cell under the current scale
is 75. That is, downlink power cannot meet the requirement. To meet such a
requirement, add some BSs.
Obtain the following table through successive iterative calculation:
BS Quantity

Required Channel Quantity

Provided Channel Quantity

83

76

71

84

76

71

85

76

71

86

76

71

87

76

71

88

65

71

If there are 88 BSs, the uplink and downlink coverage capacity requirement can be

32

met.
In the case, the BS coverage radius is

33

40.8/ 88 /1.95 0.488 Km

Vous aimerez peut-être aussi