IEICE TRANS. INF. & SYST., VOL.E98–D, NO.

7 JULY 2015
1381

LETTER
Outage Performance of MIMO Multihop Relay Network with
MRT/RAS Scheme
Xinjie WANG† , Yuzhen HUANG†† , Yansheng LI† , Nonmembers, and Zhe-Ming LU†††a) , Member

SUMMARY In this Letter, we investigate the outage performance of tivated by this important observation, several works have
MIMO amplify-and-forward (AF) multihop relay networks with maximum been devoted to investigate the impact of CCI on the perfor-
ratio transmission/receiver antenna selection (MRT/RAS) over Nakagami-
mance of relay networks [7]–[9]. To the best of our knowl-
m fading channels in the presence of co-channel interference (CCI) or not.
In particular, the lower bounds for the outage probability of MIMO AF edge, the performance of MIMO multihop relay networks
multihop relay networks with/without CCI are derived, which provides an with maximum ratio transmission/receiver antenna selection
efficient means to evaluate the joint effects of key system parameters, such (MRT/RAS) scheme for Nakagami-m fading environment is
as the number of antennas, the interfering power, and the severity of chan- unexplored from the analytical point of view regardless of
nel fading. In addition, the asymptotic behavior of the outage probability
is investigated, and the results reveal that the full diversity order can be
CCI assumptions. As such, to fill this important gap, we in-
achieved regardless of CCI. In addition, simulation results are provided to vestigate the outage performance of MIMO multihop relay
show the correctness of our derived analytical results. networks employing the MRT/RAS scheme with/without
key words: amplify-and-forward (AF), cochannel interference, MIMO, CCI. Specifically, we derive tight closed-form lower bounds
multihop, MIMO multihop relay network
and asymptotical analytical expressions for the outage prob-
ability of the considered system. The finding suggests that
1. Introduction the CCI has no impact on the diversity gain achieved by the
considered system.
The use of relaying technology is a reliable way to im-
prove the performance of wireless communication systems. 2. Network and Channel Models
Among various relay protocols, amplify-and-forward (AF)
and decode-and-forward (DF) have been broadly investi-
We consider an MIMO multihop relay network, which con-
gated [1]. In addition to the performance of single antenna
sists of the soure node S (R0 ) equipped with L0 antennas,
relay network has been analyzed under noise-limited sce-
N − 1 relays Rn , n = 1, 2, . . . , N − 1, each equipped with Ln
nario widely, e.g. [2]–[4].
antennas, and the destination D (RN ) equipped with LN an-
In an effort to improve the performance of the relay sys-
tennas. The transmission between R0 and RN is done with
tems, the multiple-input multiple-output (MIMO) transmis-
the help of N − 1 half-duplex relays Rn . The medium-access
sion technology has been integrated into relay systems [5].
control scheme allocates a frequency band to the source
Considering the increasing complexity of using multiple an-
for its transmission, which is further divided into orthogo-
tennas, antenna selection has been proposed to reduce the
nal subchannels across time using a time-division scheme
cost without affecting the achievable diversity gain. For in-
to permit half-duplex operation at the relays. Only one node
stance, the performance of joint relay and antenna selection
transmits in each time slot. In the nth time slot, Rn receives a
has been analyzed in [6] and the references therein.
faded noisy signal from the immediately preceding transmit-
However, the above works assume an ideal noise-
ting terminal Rn−1 , which is also corrupted by a finite num-
limited environment. In practical wireless communication
ber of external CCI Qn−1 . In the next time slot, Rn processes
systems, the relay transmission will inevitably be subjected
the received signal and then forwards it to Rn+1 . MRT/RAS
to co-channel interference (CCI) due to the aggressive reuse
is adopted in each hop, thus the signal received from Rn−1 at
of frequency channels for high spectrum utilization. Mo-
the k-th antenna of Rn can be expressed as
Manuscript received May 22, 2014.
Manuscript revised November 8, 2014.  
Qn

Manuscript publicized April 20, 2015. yn,k = Pn−1 h†n,k wn−1 xn−1 + Pn,q fn,k,q mn,q + nk (1)
† q=1
The authors are with the College of Communication and Elec-
tronic Engineering, Qingdao Technological University, Qingdao
266033, China. where Pn−1 is the transmit power of Rn−1 , mn,q is the q-th
††
The author is with the College of Communications Engineer- interfering signal with an average power Pn,q at Rn , and Qn
ing, PLA University of Science and Technology, Nanjing 211101, is the number of interferes at Rn . hn,k is the Ln−1 × 1 chan-
China. nel link vector between Rn−1 and the k-th receiver antenna
†††
The author is with the School of Aeronautics and Astronau-
tics, Zhejiang University, Hangzhou 310027, China.
of Rn , and its entries follow independent and identically
a) E-mail: zheminglu@zju.edu.cn distributed (i.i.d.) Nakagami-m distribution with parameters
h
DOI: 10.1587/transinf.2014EDL8102 (mhn , Ωhn ). wn−1 = hn,k is the Ln−1 × 1 weight vector,
 n,k 
Copyright 
c 2015 The Institute of Electronics, Information and Communication Engineers
 IEICE TRANS. INF. & SYST., VOL.E98–D, NO.7 JULY 2015
1382

and fn,k,q is the channel coefficient between the q-th inter- AF relay networks without CCI is lower bounded by
fering and the k-th receiver antenna of Rn , which follows  
Nakagami-m distribution with parameters (m fn , Ω fn ). nk is Pr γup ≤ γ
the AWGN at Rn with E[|nk |2 ] = σ2 , and xn−1 = Gn−1 yn−1,k N ⎡  θ  ⎤
⎢⎢⎢  γ nwv αLnwv γ ⎥⎥⎥
is the signal relayed from Rn−1 , Gn−1 is given based on AF =1− ⎣⎢ L βLnwv exp − ⎦⎥
relay protocol [8] n=1
nwv γ̄Rn γ̄Rn
⎧ N ⎡  θ ⎤
⎛  ⎞−1     ⎪ ⎪
⎨ ⎢⎢⎢ 1 nwv ⎥⎥⎥
⎜⎜⎜


Q
  ⎟⎟⎟ =1− ··· ··· ⎪
⎪ ⎢ β
⎣ Lnwv ⎥⎦ (8)
⎩
n−1
⎜ ⎟
Gn−1 = ⎜⎜⎜⎜ Pn−2

hn−1,k

F + Pn−1,q  fn−1,k,q  + σ2 ⎟⎟⎟⎟ γ̄Rn

2 2
L1wv Lnwv LNwv n=1
⎝ ⎠ ⎛ N ⎞⎫
q=1
N ⎜⎜⎜  αLnwv ⎟⎟⎟⎪ ⎪
⎬
(2) ×γ n=1 exp ⎜⎜⎝−
θnwv
γ⎟⎟⎠⎪
⎪
γ̄Rn
⎭
n=1
Based on the above model, the best antenna can be se-   Ln αLnwv wv−1n wmhn Ln−1 −2
lected based on the following equation where  Lnwv = αLnwv =1 w1 =0 ··· wvn =0 ··· wmhn Ln−1 −1 =0 ,
& '
(−1)αLnwv −1 C LnLnwv αLnwv ! %mhn Ln−1 −1
α

k∗ = arg max γn,k (3) βLnwv = wmhn Ln−1 −1 ! × vn =1 1

(wv−1n −wvn )!(vn !)wvn −wv+1n
,
k=1,2,...,Ln
mhn Ln−1 −1


2  αLnwv = 1, · · · , Ln and θnwv = vn =1 wvn with wvn =
where γn,k = Pn−1

hn,k

F σ2 . 0, 1, 2, . . . , wv−1n , vn = 1, . . . , mhn Ln−1 − 1, and vmhn Ln−1 = 0.

According to Eqs. (1)–(3), the end-to-end (e2e) SINR Proof 1: When the considered system is free of CCI, we
of MIMO multihop AF relay networks with MRT/RAS is have γneff = γn . Hence, to obtain the CDF of γend , we first
⎡ N ⎛  Qn ⎞ ⎤−1 give the CDF of γn,k as follows
⎢⎢⎢ ⎜⎜⎜ 1 + q=1 γIqn ⎟⎟⎟ ⎥⎥
γend = ⎢⎢⎣ ⎢ ⎜
⎜⎜⎝1 + ⎟⎟⎠ − 1⎥⎥⎥⎥⎦
⎟ (4)  
γn Υ mhn Ln−1 , γ̄Rx
n=1
Fγn,k (y) = ( )n (9)
 2  Γ mhn Ln−1
where γn = γn,k∗ , and γIqn = Pn,q  fn,k∗ ,q  σ2 is the q-th
where γ̄Rn = Pn−1 Ωhn /mhn σ2 , Υ (·, ·) is the lower incomplete
interference-to-noise ratio (INR) at the selected Rn antenna.
Gamma function [10, Eq. (8.350.1)] , and Γ (·) is the Gamma
In general, an exact analysis of the statistics of e2e
function [10, Eq. (8.310.1)] . According to Eq. (3) with [11,
SINR in Eq. (4) is very challenging. Hence, to circumvent
Eq. (9)] , the CDF of γn can be derived as
this difficulty, we adopt the following tight upper bound [8]
γ

Ln Υ(m L
hn n−1 , γ̄R )
γend ≤ γup = min(γ1eff , γ2eff , . . . , γeff
N ) (5) Fγn (γ) = n

 k=1
Γ(mhn Ln−1 )
Qn
where γneff = γn /(γI n + 1) and γI n = q=1 γIqn .   θ  
 γ nwv αLnwv γ
=1− βLnwv exp − (10)
3. Outage Performance Analysis L
γ̄Rn γ̄Rn
nwv

As an important performance indicator for wireless commu- Finally, substituting Eq. (10) into Eq. (7), the exact analyti-
nication systems, the outage probability can be expressed as cal expression for the lower outage bound of the considered
system without CCI can be obtained as Eq. (8) after some
Pout (γ) = Pr (γend ≤ γ) = Fγend (γ) (6) simple mathematical manipulations.
Based on Eq. (5), it is lower bounded by To gain further insights, we now look into the high
SNR regime, and derive the asymptotic expression for the

N   outage probability.
Fγup (γ) = 1 − 1 − Fγneff (γ) (7)
n=1 Corollary 1: When the considered system is free of CCI,
the asymptotic outage probability is given by
To evaluate the impact of CCI on the performance of MIMO
 
multihop AF relay systems with MRT/RAS, the following Pr γup ≤ γ
two cases of interest are considered, i.e. MIMO multihop AF
N ⎪
⎧  m L L ⎫
relay systems without CCI and MIMO multihop AF relay  ⎪
⎨ γ hn n−1 n 1 ⎪
⎪
⎬
≈1− ⎪
⎪ 1 − * + Ln ⎪
⎪
systems with CCI. ⎩ γ̄Rn (mhn Ln−1 )! ⎭
n=1
( )−Gdnc
3.1 MIMO Multihop AF Relay Systems without CCI = Ganc γ̄Rn∗ (11)
, -
where n∗ = min mhn Ln−1 Ln , the outage diversity gain
If CCI signals do not exist in the system, we have the fol- n=1,2,...,N
lowing important results. Gdnc = mhn∗ Ln∗ −1 Ln∗ , and the outage array gain Ganc =
. * +Ln∗ −1 /− Gdnc −1
1

Theorem 1: The outage probability of MIMO multihop n∗ (mhn∗ Ln∗ −1 )! γ .

LETTER
1383

 θnwv   θ
Proof 2: To derive the asymptotical CDF of Fγup (γ), we  αLnwv γ γ nwv
first express the incomplete Gamma function using the se- =1− ηLnwv exp −
Lnwv np=0
γ̄Rn γ̄Rn
ries expansion as [10, Eq. (8.354.1)]  γ̄ np+m fn Qn
Rn

∞
(−1)n (θx)n (θx)m αLnwv
Υ (m, θx) = (θx)m ≈
x→0
. (12) ×Γ(np + m fn Qn ) . /np+m fn Qn (16)
n=0
n! (m + n) m γ+
γ̄Rn
αLnwv γ In

Then, substituting Eq. (12) into Eq. (10), the CDF of Fγn (γ)
Now, by substituting Eq. (16) into Eq. (7), the lower
can be approximated as
bound for the outage probability of the considered system
 m L L with CCI can be obtained as Eq. (14) after some algebraic
y hn n−1 n 1
Fγ∞n (y) ≈ * + . (13) manipulations.
γ̄Rn (mhn Ln−1 )! Ln
Note that although the closed-form expression in
To this end, by substituting Eq. (13) into Eq. (7), we have the
Eq. (14) provides an efficient means to evaluate the outage
asymptotic outage probability as shown in Eq. (11).
probability of the system at arbitrary SINRs, the expression
is in general too complex to yield any insights. Motivated
3.2 MIMO Multihop AF Relay Systems with CCI by this, we hereafter pursue an asymptotic analysis in the
high SNR regime and we have the following key results.
If CCI signals are considered, we have the following key
Corollary 2: When CCI are considered, the asymptotic
result.
outage probability is given by
Theorem 2: The outage probability of MIMO multihop
N ⎪
⎧
AF relay networks with MRT/RAS in the presence of CCI    ⎪
⎨ 1
Pr γup ≤ γ = 1 − ⎪
⎪ 1− * +
is lower bounded by ⎩ (mhn Ln−1 )! Ln
n=1
⎧ N
θNwv ⎪    ⎫
   
  θ1wv  ⎪
⎨
mhnLn−1 Ln
Γ(tn + m fn Qn ) γ mhn Ln−1 Ln ⎪
⎪
Pr γup ≤ γ = 1 − ··· ··· ⎬
⎪
⎪
⎩ × tn tn
γIn Cmh Ln−1 Ln ⎪
⎪
L1wv LNwv 1p=0 N p=0 n=1
n Γ(m f Qn ) γ̄R ⎭
tn =0 n n
⎡  ⎤ ⎛ N ⎞  −Gd
⎢⎢⎢ ηLnwv Γ np + m fn Qn ⎥⎥⎥ ⎜⎜⎜  αLnwv ⎟⎟⎟ = Ga γ̄Rn‡ (17)
× ⎢⎣⎢ ⎢ ⎥
⎥⎦⎥ exp ⎜⎜⎝− γ⎟⎟
γ̄Rθnwv γ̄Rn ⎠ , -
where n‡ =
n=1
n
⎡ ⎤ min mhn Ln−1 Ln , the outage diversity gain
N ⎢
⎢⎢  γ̄Rn np+m fn Qn ⎥⎥⎥ ⎫ n=1,2,...,N
N  ⎢⎢⎢ ⎥⎥⎥ ⎪ ⎪ Gd = mhn‡ Ln‡ −1 Ln‡ n‡ , and the outage array gain Ga =
×γ n=1 θnwv ⎢⎢⎢ . αLnwv / ⎥⎥ ⎬
np+m fn Qn ⎥ ⎥⎦⎥ ⎪
⎢⎢⎢ ⎥ ⎪ (14) ⎡ ⎤− 1
⎭ t t
⎢⎢⎢ mh ‡ Ln‡ −1 Ln‡ Cmnh‡ Ln‡ −1 Ln‡ γInn‡‡ Γ(tn‡ +m fn‡ Qn‡ ) ⎥⎥⎥ Gd
n=1 ⎣ γ + γ̄Rn
αLnwv γ ⎢⎢⎢ ‡ n
&.n
‡
/ 'L ‡ ⎥⎥⎥ γ−1 .
⎣ n tn‡ =0 ⎦
In
n
np
m Lh ‡ !
n‡ −1 Γ(m Q )f ‡ n‡
βLnwv Cθnwv n n
where γ In = PIqn Ω fn /m fn σ2 and ηLnwv = m f Qn .
Γ(m fn Qn )γ In n Proof 4: The result can be obtained by following similar
lines as in the proof of Corollary 1, along with some simple
Proof 3: According to [7], the PDF of γI n is expressed as
algebraic manipulations.
 
zm fn Qn −1 z
fγIn (z) = m Q   exp − (15) The main insight observed from Corollaries 1 and 2 is
γ fn n m f Qn − 1 ! γ In that the diversity order achieved by MIMO multiple hop AF
In n

relay networks with/without CCI is completely decided by

By invoking the concepts of probability theory, the the number of antennas and the channel fading severity, CCI
CDF of Fγneff (γ) can be written as only has impact on the outage array gain.
 
γn
Fγneff (γ) = Pr <γ 4. Numerical Results and Discussion
γI n + 1
0 ∞ 1 2θ  
 (z + 1) γ nwv αLnwv γ In this Section, the Monte Carlo simulations are provided
=1− βLnwv exp −
0 L γ̄Rn γ̄Rn to validate the derived analytical expressions. Without loss
nwv
  of generality, the transmit power of all nodes Rn is assumed
zm fn Qn −1 z αLnwv γz
× m Q   exp − − dz to be equal, i.e., P s , the channel mean powers are given by
γIn fn n m fn Qn − 1 ! γIn γ̄Rn Ωhn = 1 and Ω fn = 1, and the SNR threshold γ is 5dB.
 θnwv  θ 0 With the aim to highlight the joint effect of the number
 γ nwv ∞ np+m fn Qn −1 of antennas and the channel fading severity parameters on
=1− ηLnwv z
Lnwv np=0
γ̄Rn 0 the outage performance, we denote the combination param-
  eters as {L0 , . . . , LN ; mh1 , . . . , mhN }. Three different repre-
αLnwv γ (z + 1) z
× exp − − dz sentative examples are considered: S 11 = {1, 1, 1, 1; 1, 1, 1},
γ̄Rn γ In

1384
Fig. 1 Outage probability of three hop MIMO relay networks without
CCI.
S 12 = {1, 1, 1, 1; 2, 2, 2}, S 13 = {2, 2, 2, 2; 1, 1, 1}. Figure 1
plots the outage probability versus P s for three hop MIMO
relay networks without CCI. We can see that, the SNR upper
bound in Eq. (5) is tight when SNR increases. The analyti-
cal results are in exact agreement with the simulation curves,
and the asymptotic curves tight converge to the exact curves
in the high SNR regime, which verifies the accuracy of our
analytical derivation. In addition, as can be expected, the
outage probability decreases as the number of transmitter
and receiver antennas increases, e.g., S 13 can achieve better
performance than S 11 . Furthermore, as the diversity order is
governed by the minimum , of mhn Ln−1
- Ln for each hops, the
higher the value of min mhn Ln−1 Ln , the better the outage
performance. Figure 2 plots the outage probability versus
P s for three hop MIMO relay networks in the presence of
two i.i.d interferers. The channel fading severity parame-
ters between CCI and Rn is m fn = 2. Two kinds of differ-
ent interfering power are considered. i.e., Pn,q = 1dB and
Pn,q = 5dB. As can be expected, the outage performance
can be significantly improved when the fading severity mhn
increases, which indicates the outage probability is closely
related to the channel quality of each link. However, in-
creasing the interfering power will decrease the outage per-
formance, while we can improve the outage probability of
the system by increasing the number of antennas at each
nodes, since the extra number of antennas provide the addi-
tional outage array gain to the system.
5. Conclusion
This Letter has studied the outage performance of MIMO
multihop AF relay networks with MRT/RAS scheme. The
closed-form analytical expressions for the outage probabil-
ity of the considered system with/without CCI were derived,
they provide a fast and efficient means of evaluating system
performance. Moreover, simple and informative high SNR
approximations for the outage probability were provided,
they enable us to investigate the impact of the number of
antennas at each node, channel fading severity parameters,
and the interfering power on the outage performance. Our
IEICE TRANS. INF. & SYST., VOL.E98–D, NO.7 JULY 2015
Fig. 2 Outage probability of three hop MIMO relay networks with CCI.
finding reveals that CCI has no impact on the outage diver-
sity gain, and it only affects the outage array gain achieved
by the considered system.
Acknowledgments
This work is partially supported by Shandong Provincial
Natural Science Foundation of China (No. ZR2010FM011)
and Zhejiang Provincial Natural Science Foundation of
China (No. R1110006).
References
[1] J.N. Laneman, D. Tse, and G.W. Wornell, “Cooperative diversity in
wireless networks: Efficient protocols and outage behavior,” IEEE
Trans. Inf. Theory, vol.50, no.12, pp.3062–3080, Dec. 2004.
[2] S. Ikki and M.H. Ahmed, “Performance analysis of cooperative di-
versity wireless networks over Nakagami-m fading channel,” IEEE
Commun. Lett., vol.11, no.4, pp.334–336, April 2007.
[3] M.O. Hasna and M.-S. Alouini, “Outage probability of multi-
hop transmission over Nakagami fading channels,” IEEE Commun.
Lett., vol.7, no.5, pp.216–218, May 2003.
[4] V. Asghari, A. Maaref, and S. Aissa, “Symbol error probability anal-
ysis for multihop relaying over Nakagami fading channels,” in Proc.
IEEE WCNC 2010, Sydney, Australia, pp.1–6, April 2010.
[5] Y. Fan and J.S. Thompson, “On the outage capacity of MIMO
multihop networks,” in Proc. IEEE GlOBECOM, St. Louis, MO,
pp.2204–2208, Dec. 2005.
[6] G. Amarasuriya, C. Tellambura, and M. Ardakani, “Joint relay
and antenna selection for dual-hop amplify-and-forward MIMO
relay networks,” IEEE Trans. Wireless Commun., vol.11, no.2,
pp.493–499, Nov. 2012.
[7] F.S. Al-Qahtani, T.Q. Duong, C. Zhong, K.A. Qaraqe, and H.
Alnuweiri, “Performance analysis of dual-hop af systems with in-
terference in Nakagami-m fading channels,” IEEE Signal Process.
Lett., vol.18, no.8, pp.454–457, Aug. 2011.
[8] S.S. Ikki and S. Aissa, “Multihop wireless relaying systems in
the presence of cochannel interferences: Performance analysis and
design optimization,” IEEE Trans. Veh. Technol., vol.61, no.2,
pp.566–573, Feb. 2012.
[9] Y. Huang, F. Al-Qahtani, C. Zhong, Q. Wu, J. Wang, and H.
Alnuweiri, “Performance analysis of multiuser multiple antenna re-
laying networks with co-channel interference and feedback delay,”
IEEE Trans. Commun., vol.62, no.1, pp.59–73, Jan. 2014.
[10] I.S. Gradshteyn and I.M. Ryzhik, Table of Integrals, Series, and
Products, 7th ed., Academic, San Diego, CA, 2007.
LETTER
1385

[11] S. Choi and Y.-C. Ko, “Performance of selection MIMO systems

with generalized selection criterion over Nakagami-m fading chan-
nels,” IEICE Trans. Commun., vol.E89-B, no.12, pp.3467–3470,
Dec. 2006.