Vous êtes sur la page 1sur 3

410

IEEE COMMUNICATIONS LETTERS, VOL. 15, NO. 4, APRIL 2011

Outage Analysis for Multiuser Two-Way Relaying in


Mixed Rayleigh and Rician Fading
Haiyang Ding, Jianhua Ge, Daniel Benevides da Costa, Member, IEEE, and Yi Guo
AbstractThis Letter analyzes the outage performance of a
multiuser bidirectional cooperative cellular network in which
one base station (BS) and one out of mobile stations (MSs)
exchange messages with the aid of one amplify-and-forward (AF)
relay. Our analysis allows for a practical fading scenario where
the MS-relay and the BS-relay channels undergo independent
Rayleigh and Rician fading, respectively. In this case, a tight
closed-form lower bound for the outage probability of the
considered system is derived. Furthermore, asymptotic analysis is
performed to characterize the outage behavior in high signal-tonoise ratio (SNR) regime. Interestingly, in high SNR regime, it
is analytically shown that when the number of MS is greater
than one, the outage performance is dominated by the BSrelay link regardless of the number of MS and of the fading
severity pertaining to MS-relay links. Finally, simulation results
are presented in order to corroborate the theoretical analysis.

fading1 , the outage performance of a multiuser two-way


relaying system is studied. Particularly, a tight closed-form
lower bound is derived for the outage probability of the considered system, from which asymptotic analysis is performed
to characterize the outage behavior at high signal-to-noise ratio
(SNR). Simulation results are presented to corroborate the proposed methodology. Throughout this Letter, () and ()
denote the probability density function (PDF) and cumulative
distribution function (CDF) of a variate , respectively.

Index TermsAmplify-and-forward relays, bidirectional cooperative networks, multiuser scheduling.

Consider a bidirectional cooperative cellular network where


one BS intends to exchange information with one out of
mobile stations MS ( = 1, 2, ..., ) with the aid of one AF
relay R. All terminals are single-antenna devices and operate
in a half-duplex mode. Due to the presence of obstacles, it
is assumed that there is no direct link between the MSs and
BS. For multiuser diversity (MUD), at a time, the best MS
(e.g., MS ) is selected from the available ones and the
specific selection criterion will be addressed later. Afterwards,
a two-phase bidirectional transmission starts [5]. In the first
phase, BS and MS transmit their respective signals to R
simultaneously. In the second phase, R amplifies the received
signals with a variable gain [5] and forwards it to both BS and
MS . Finally, self-interferences are suppressed, respectively,
at BS and MS . Skipping the detailed signal transmission
process (please, refer to [5, Eqs. (1)-(5)] for details), the endto-end SNRs pertaining to the links BS R MS and
MS R BS can be expressed, respectively, as

I. I NTRODUCTION

OWADAYS, in order to pursue the potential spectrum


efficiency enhancement, there has been a growing interest in the design and performance analysis of two-way
relaying systems [1][4]. More specifically, assuming that
one [1], [2] or multiple [3], [4] relays were available, the
authors of [1][4] studied the average sum rates or outage
performance for several two-way relaying systems with two
users. However, in practice, multiple potential users may
compete to perform bidirectional communications. Adopting
this practical concern, very recently [5] the authors proposed
a multiuser two-way relaying framework where one base
station (BS) and one selected mobile station (MS), among
multiple potential ones, exchange messages with the aid of one
amplify-and-forward (AF) relay. Based on this framework, the
average rates were analyzed in [5]. On the other hand, another
important performance measure, i.e., outage probability, which
characterizes the transmission robustness, remains unknown
for such systems. Besides, unlike the MS-to-relay channels,
it may be inappropriate to model the BS-to-relay channel as
Rayleigh fading (as assumed in [5]) due to a strong line-ofsight (LOS) component [6], [7]. In such a case, Rician fading
is often employed to model these channels [6][8].
In this Letter, assuming the BS-to-relay and the relay-toMS channels are, respectively, subject to Rician and Rayleigh
Manuscript received November 17, 2010. The associate editor coordinating
the review of this letter and approving it for publication was C. Comaniciu.
H. Ding, J. Ge, and Y. Guo are with the State Key Laboratory of Integrated
Service Networks, Xidian University, Xian, China. H. Ding is also with Xian
Communication Institute, Xian, China (e-mail: dinghy2003@hotmail.com,
jhge@xidian.edu.cn, guoyi402@163.com). This work is supported by the
National Natural Science Foundation of China (Grant No. 61001128 and
61001207).
D. B. da Costa is with the Federal University of Ceara - Campus Sobral,
CE, Brazil (e-mail: danielbcosta@ieee.org). This work is supported by the
FUNCAP (Grant No. BP1-0031-00090.01.00/10).
Digital Object Identifier 10.1109/LCOMM.2011.022411.102238

II. S YSTEM AND C HANNEL M ODELS

1, =

1 1
0

1 1 3 2,
0
0
( + )
+ 2 30 2,

+1

, 2, =

3 1 2 2,
0
0
2 2,
(1 +3 )1
+
0
0

+1
(1)

where 1 , 2 , and 3 denote the transmit powers of BS,


MS , and R, respectively, 0 stands for the mean power
of the additive white Gaussian noise (AWGN) arriving at
each terminal in the system2 and, 1 and 2, indicate the
channel gains pertaining to the links BS R and MS R,
respectively. As in [5], perfect channel reciprocity is assumed.
Concerning to the channel fading characteristic, we assume
the link between BS and R is subject to Rician fading owing
to one strong direct LOS component [6][8], whereas the
links between R and MS are subject to independent but not
1 It is worthwhile to mention that this mixed Rayleigh/Rician propagation
characteristic is well-documented in a recently released WINNER II project
deliverable in micro/macro cellular multi-hop transmissions [7].
2 Note that the authors of [5] normalized to unity the mean power (variance)
of the AWGN arriving at each terminal, whereas in this work we do not
impose such constraint. Following the same procedure as [5], (1) is readily
attained.

c 2011 IEEE
1089-7798/11$25.00

DING et al.: OUTAGE ANALYSIS FOR MULTIUSER TWO-WAY RELAYING IN MIXED RAYLEIGH AND RICIAN FADING

necessarily identically distributed (i.n.i.d) Rayleigh fading3 .


Accordingly, the PDF of 1 conforms to noncentral chisquare distribution given
( by [8, Eq.)(2.16)], i.e., 1 () =
(1+)
(1+)
1

0 2 (1+)
, with being the Ri1
1
cian -factor, 1 denoting the mean value of 1 , and 0 ()
denoting the zeroth-order modified Bessel function of the first
kind [9, pp. 916]. In addition, the channel gains 2, conform
to exponential distribution with mean 2, .
In this way, the transmission rates pertaining to the links
BS R MS and MS R BS can be expressed
as 1, 12 log2 (1 + 1, ) and 2, 12 log2 (1 + 2, ),
respectively. Interestingly, according to [5], the MS with the
max } [2, ].
maximum 1, and 2, satisfies to = arg{1,...,
Therefore, for the MUD-based MS selection, MS is naturally chosen to perform the two-way transmission toward BS,
yielding the transmission rates 1, 12 log2 (1 + 1, ) and
2, 12 log2 (1 + 2, ), respectively.
III. O UTAGE P ROBABILITY A NALYSIS
For a predefined threshold rate bit/s/Hz, the outage
probability is defined as the probability that either 1, or
2, falls below , which can be formulated as
(
)
out = Pr min [1, , 2, ] < 22 1
= 1 Pr (1, > , 2, > ) ,

(2)

in which Pr() indicates probability. For ease of notation, we


define /0 ( = 1, 2, 3), leading to (3), given at the
top of the next page, in which 2 {1,...,
max } [2, ] and step (a)
becomes extremely tight for high SNR ( 1), as shown
in [1][3]. In order to facilitate the subsequent analysis, the
following lemma is presented.
Lemma 1: The PDF and the CDF of 1/2 can be
written as

() = 2
(1)+1 ,
=1 {1,..., }

(4a)

() =

=1

{1,..., }

+1

(1)

(4b)

( = 1, 2, 3) [1][3]. Then, can be rewritten as


(
) 

= Pr <
2, <

() ()
=

2
2

3
3
2 2
2 2
(, ) +
(, ),
=
0

where
Proof : From [10, Theorem 1] and invoking the Fundamental
Theorem given in [11, Sec. 5-2], the PDF of can be derived
as (4a). Then, by integrating (4a) with respect to , the CDF
of can be attained as (4b).

Similarly, from [11, Sec. 5-2], the PDF of = 1/1 is


obtained and its CDF can be readily attained by using [12, Eq.
(9)]. In what follows, the term in (3) will be determined.
For such, without any loss of generality, we assume4 =
3 In this Letter, we assume the R MS links ( = 1, ..., ) are subject

to i.n.i.d Rayleigh fading, whereas [5] assumed independent and identically


distributed (i.i.d) Rayleigh fading for these links. Besides, since herein the
BS R link is subject to Rician fading, the system model in [5] is indeed
a special case of the counterpart considered in this work.
4 Herein, similar to [13], we define as the system SNR since it is
proportional to the transmit and receive SNR at each terminal in the system.

(5)

where (, ) () (). Unfortunately, closed-form


solutions for (5) are not available, which makes it quite
difficult to continue the analysis. Nevertheless, by enlarging
slightly the integration regions of (5), can be upper
bounded by


2
2

(, ) +
(, )
0

() () +


0

() () .

(6)

Next, both 1 and 2 will be addressed. By substituting


the PDF of and (4b) into 1 and after some algebraic
manipulations, it follows that
1 =

=1

{1,..., }

(1)+1

1+
1

1,

0 (),

(7)

where (1+)
and 2 (+1)
1
1 . Now, making
the change of variates and using [12, Eq. (9)], a closed-form
expression for 1, can be attained as
/2

(
)

+
1, =
( + ) , (8)

,2
+
+

where 2 /2, / , and (, ) denotes the


first-order Marcum -function [8, Eq. (4.34)]. Turning our
attention to 2 , by plugging the CDF of and (4a) into 2
and after some arrangements, we arrive at
2 =

=1

{1,..., }

1
2, .

411

(1)+1

)
(

1
2

2,
.
2

(9)

2,

Then, applying the change of variates and using [12, Eq. (40)],
2, can be calculated as
)
(
[

2, =
2, 2

(
)]


( + ) .
+
,2
+

(10)
Now, we summarize the above analytical results in the following proposition.
Proposition 1: For the multiuser two-way relaying system
under study, a closed-form lower bound for the outage prob-

412

IEEE COMMUNICATIONS LETTERS, VOL. 15, NO. 4, APRIL 2011

()

out 1 Pr

)
1 3 1 2
2 3 1 2
> ,
> ,
(2 + 3 )2 + 1 1
(1 + 3 )1 + 2 2

10

ability can be formulated as

=1

{1,..., }

=1

{1,..., }

+1

(1)

1,

LB
(1)+1 2, out
,

(11)

Proof : For sufficiently high SNR, i.e., 1, by applying


the Taylor series expansion of (, ) in (11) [8, Eq. (4.41)]
and with the aid of [9, Eqs. (9.6.7) and (3.1.1)], it follows that

2
2
+

(1)+1 .

=1 {1,..., }
=

K=0 (Rayleigh)
N=3, = =1
1

K=0dB
N=1,1=1, 2=2

where 1, and 2, are determined by (8) and (10), respectively.

LB
Proposition 2: For sufficiently high SNR, i.e., 1, out
can be asymptotically expressed as
(
)
{
2
1

+
LB

2 , if = 1
(12)

out
2
, if 2

LB
out

K=0dB
N=1,1=2=1

10

Outage probability

out 1

(3)

(13)

10

K=0dB, N=3
= =1

K=5dB, N=3
= =1

K=7.5dB, = =1

10

N=3 (black), 5 (red)

10

Simulation
Lower bound, (11)
Asymptote, (12)

K=7.5dB, =2
1
N=3, =1 (black)
2
N=5, =2 (red)
2

10

Fig. 1.

10

15
20
System SNR (E), dB

25

30

35

Outage probability versus system SNR ().

cases, the asymptotic curves attained from (12) and the


corresponding simulated ones overlap in high SNR regions,
which corroborates the proposed asymptotic analysis. It is
worth noting that for 2 and for a given Rician factor, e.g. = 7.5 dB, the outage performance does not
vary with the number of MSs ( ) or with the fading severity
of the MS-relay links (2 ) at high SNR, validating therefore
the Remarks. However, for = 1, the outage performance is
determined by both hops, as expected.

For = 1, we have = 1/2 . On the other hand, for 2,


we define a new function as below

(
R EFERENCES
)


1 2, . [1] Y. Han, S. H. Ting, C. K. Ho, and W. H. Chin, Performance bounds
(1) =
() 1 +
=1

{1,..., }
=

=1

Accordingly, we have = ()
=0 = 0. Therefore, it
follows that 0 for 2, which completes the proof.

Remarks: From Proposition 2, it can be observed that the


diversity order of the multiuser two-way relaying system is
unity, which is due to the fact that there is only one single
link between BS and R in the considered system. Interestingly,
from (12) it is shown that when 2, the outage behavior in
high SNR regime is dominated by the BS R link regardless
of the number of MS ( ) and the fading severity of the MSrelay links (2, , = 1, . . . , ), whereas this conclusion does
not hold for = 1.
IV. N UMERICAL R ESULTS AND D ISCUSSIONS
In this Section, for illustration purposes and without loss
of generality, we assume 2, = 2 for = 1, 2, ..., .
Also, the end-to-end threshold rate is set to = 1 bit/s/Hz.
Fig. 1 plots the outage probability versus system SNR .
From this figure, it is clear that the lower bound obtained
from (11) matches very well with simulations in the medium
and high SNR regions. With an increase of the Rician factor pertaining to the BS R link, the outage performance
improves considerably, as expected. In addition, for all the

[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]

for two-way amplify-and-forward relaying, IEEE Trans. Wireless Commun., vol. 8, no. 1, pp. 432439, Jan. 2009.
Q. Li, S. H. Ting, A. Pandharipande, and Y. Han, Adaptive two-way
relaying and outage analysis, IEEE Trans. Wireless Commun., vol. 8,
no. 6, pp. 32883299, June 2009.
H. Guo and J. H. Ge, Outage probability of two-way opportunistic
amplify-and-forward relaying, IET Electron. Lett., vol. 46, no. 13, pp.
918919, June 2010.
H. Ding, J. Ge, D. B. da Costa, and Z. Jiang, Outage performance of
fixed-gain bidirectional opportunistic relaying in Nakagami-m fading,
IET Electron. Lett., vol. 46, no. 18, pp. 12971299, Sep. 2010.
Y.-U. Jang and Y. H. Lee, Performance analysis of user selection for
multiuser two-way amplify-and-forward relay, IEEE Commun. Lett.,
vol. 14, no. 11, pp. 10861088, Nov. 2010.
H. Suraweera et al., Two hop amplify-and-forward transmission in
mixed Rayleigh and Rician fading channels, IEEE Commun. Lett.,
vol. 13, no. 4, pp. 227229, Apr. 2009.
P. Kyosti, J. Meinila, and L. Hentila et al., WINNER II Interim
Channel Models (IST-4-027756 WINNER II D1.1.1 V1.1). Available:
http://www.istwinner.org/WINNER2-Deliverables/D1.1.1.pdf.
M. K. Simon and M. -S. Alouini, Digital Communication over Fading
Channels, 2nd edition. John Wiley & Sons, 2005.
M. Abramowitz and I. A. Stegun, Handbook of Mathematical Functions
with Formulas, Graphs, and Mathematical Tables. New York, 1972.
A. Bletsas, A. G. Dimitriou, and J. N. Sahalos, Interference-limited
opportunistic relaying with reactive sensing, IEEE Trans. Wireless
Commun., vol. 9, no. 1, pp. 1420, Jan. 2010.
A. Papoulis, Probability, Random Variables, and Stochastic Processes,
4th edition. McGraw-Hill, 2002.
A. H. Nuttall, Some integrals involving the Q-function, Naval Underwater Systems Center (NUSC), Technical Report 4297, Apr. 1972.
Y. Zhao, R. Adve, and T. J. Lim, Improving amplify-and-forward
relay networks: optimal power allocation versus selection, IEEE Trans.
Wireless Commun., vol. 6, no. 8, pp. 31143123, Aug. 2007.

Vous aimerez peut-être aussi