Documente Academic
Documente Profesional
Documente Cultură
DOI 10.1007/s11277-015-2901-5
1 Introduction
123
2234 S. S. Chauhan, S. Kumar
Antenna selection approach is an efficient and practical solution to overcome the cost
and hardware complexity of MIMO system while preserving the benefits of all antennas, is
proposed in [3–7]. The primary goal behind antenna selection is to prefer only the optimal
set of antennas at both transmit and/or receive ends in MIMO systems to maximize the
spectral efficiency and improve error performance. The upper bounds for MIMO system
capacity has been obtained by selecting the excellent subset of antenna at one link end in
[3]. The transmit antenna selection (TAS) was applied in [4] and the symbol error rate
(SER) has been investigated for M-ary signal. The error performance analysis of TAS/
MRC was evaluated over Nakagami-m fading channels in [5]. In [6], authors has been
examined the SER expression for binary and M-ary modulations over an independent and
non-identically distributed (i.n.d.) Nakagami fading channels. Authors in [7] has been
presented the TAS with switch-and-examine (SEC) system and outage probability, average
received SNR per branch, and SER expressions for different modulation schemes over an
i.n.d. Rayleigh fading has been derived. It has been noticed that with the estimation of
fewer branches, the performance of TAS/SEC system is superior compared to that of TAS/
SC system. The capacity of fading channels varies which is based on the availability of
channel state information at the receiver (CSIR) and at the transmitter (CSIT) of the
wireless link. The CSIT techniques are also known as adaptive transmission (AT) tech-
niques. The CSI estimated at the receiver consist of the signal-to-noise ratio (SNR) level
and is fed back to the transmitter. Then, AT can be operated on one frame to other frame so
that permitting for distinct adjustments among the attainable channel capacity and the
corresponding implementation complexity. In [8], authors have been evaluated average
channel capacity of diversity-combining techniques under CSIT techniques over Rayleigh
fading channel. Authors in [9], has been derived the capacity of orthogonal space–time
block code (OSTBC) in Rayleigh fading under distinct AT and estimation error. The
capacity of log-normal channels has been analytical derived in [10] over distinct AT
techniques. Authors in [11] introduced asymptotic capacity analysis at high and low SNRs
under CSIT techniques. In [12] and [13], authors has been evaluated the capacity of
OSTBC system under various AT techniques over correlated Nakagami and Weibull
MIMO fading channels respectively.
The analytical comparison of average channel capacity of TAS/MRC with AT tech-
niques over uncorrelated Nakagami-m fading channel has not been studied. Thus, this
correspondence fills the gap by evaluating the exact channel capacity expression for TAS/
MRC under variable power and rate adaptation techniques over Nakagami-m fading
channel. Our outcomes evaluated the upper bound attainable spectral efficiency of TAS/
MRC for any practical application.
The remainder of this paper is organized as follows: First, in Sect. 2, the characteri-
zation of TAS/MRC system-model is presented. In Sect. 3, we examine the channel
capacity expression of TAS/MRC for the considered power and rate adaptation schemes.
Finally, Sect. 4 presents numerical results, followed by concluding statements in Sect. 5.
We assume a TAS/MRC (Nt, 1; Nr) system in Nakagami frequency non selective fading
channel environment with Nt transmit and Nr receive antennas. Let the knowledge of CSI is
perfectly available at the receiver and based on which receiver selects
one out of Nt
transmit antennas through an error free feedback links. Let H ¼ hij 2 CNr Nt be the
123
Adaptive-Transmission Channel Capacity of Maximal-Ratio… 2235
complex channel matrix follows the Nakagami-m distribution and the coefficient of
channel are modeled as hij CN ð0; 1Þ. The input–output relationship at any time t can be
expressed as
Y ¼ HX þ V ð1Þ
where matrix Y 2 C Nr 1 is complex received signal column vector, X 2 CNt 1 is the
complex transmitted signal column vector and V 2 CNr 1 is a receiver noise vector and are
modeled as V CN ð0; N0 Þ. We have selected a single transmit antenna for data trans-
mission at any time with largest channel gain to maximize the received signal power,
denoted by J, and is given as
( )
X
Nr
2
J ¼ arg max Lj ¼ hij ð2Þ
1 j Nt i¼1
The arrangement of random variables Lj are as LðNt Þ Lð2Þ Lð1Þ and denoted by
Lv where 1 B v B Nt. The instantaneous post processing SNR of TAS/MRC output can be
given as cNt ¼ cLðNt Þ where c is the average received SNR per channel. In a frequency flat
Nakagami-m fading, the probability density function (pdf) and cumulative density function
(cdf) is given by [5]
mNr
m 1
cmNr 1 eð c Þ
mc
fc ðcÞ ¼ ð3Þ
c CðmNr Þ
and
c mNr ; m cc
Fc ðcÞ ¼ ð4Þ
CðmNr Þ
respectively, where c(a, z) is incomplete gamma function.
The pdf of cNt is given by
N 1
fcNt ðcÞ ¼ Nt Fc ðcÞ t fc ðcÞ
mNr mNr 1
ð5Þ
m c ðmc
c Þ c mN ; m
c Nt 1
¼ Nt e r
c CðmNr ÞNt c
h iNt 1
In order to obtain the pdf of cNt in simple form, we have to expand c mNr ; m cc in
(5), by using the identity [14, 0.314]
!n
X
1 X
1
k
ak z ¼ ck z k ð6Þ
k¼0 k¼0
Pp
where c0 ¼ an0 ; cp ¼ pa10 k¼1 ðkn p þ kÞak cpk and c(b, z) is defined as [14, (8.351.2)].
ez zb
cðb; zÞ ¼ ð1; b þ 1; zÞ ð7Þ
b
in which 1F1(m; n; p) is the first kind confluent hypergeometric function [14]
123
2236 S. S. Chauhan, S. Kumar
X1
ðmÞk pk
1 F1 ðm; n; pÞ ¼ ð8Þ
k¼0
ðnÞk k!
h iNt 1
where ðaÞk ¼ CCðaþk
ðaÞ
Þ
is pochhammer’s symbol [14]. Thus c mNr ; m c
c can be written as
mNr þk
c Nt 1 X 1
c mcðNt 1Þ
c mNr ; m ¼ ck m ef c g ð9Þ
c k¼0
c
Nt 1 Pp
where c0 ¼ 1
mNr ; cp ¼ CðmNpr þ1Þ kNt p
k¼1 CðmNr þkþ1Þ cpk
Now inserting (9) into (5), the pdf of cNt is given as
X
1 kþ2mNr
Nt m mc
fcNt ðcÞ ¼ Nt c k ckþ2mNr 1 eð c Nt Þ ð10Þ
CðmNr Þ k¼0 c
In this section, the capacity of TAS/MRC over Nakagami-m MIMO fading channels under
different AT techniques is considered and the exact analytical expressions for the corre-
sponding capacities are evaluated.
In this scheme, the data rate is varied by transmitter according to the channel statistics
while keeping the transmit power constant. The average channel capacity in this scheme
Cora[b/s/Hz] is given by [9–13]
Z1
1
Cora ¼ lnð1 þ cÞfcNt ðcÞdc ð11Þ
ln2
0
Inserting (10) into (11) and utilizing the integral from [8]
Z1
J n ðvÞ ¼ lnð1 þ tÞtn1 evt dt; v [ 0; n ¼ 1; 2. . .
0 ð12Þ
X
n
Cðn þ j; vÞ
¼ ðn 1Þ!ev
j¼1
vj
emNt =c j
j¼1 mNt
c
123
Adaptive-Transmission Channel Capacity of Maximal-Ratio… 2237
In this scheme, the channel statistics must be known at both transmitter and receiver. The
average channel capacity Copra for the scheme is given by [9–13] as
Z1
c
Copra ¼ log2 fc ðcÞdc ð14Þ
c0 Nt
c0
where c0 is the optimal cutoff SNR below which data transmission is not valid and must
satisfy
Z1
1 1
fcNt ðcÞdc ¼ 1 ð15Þ
c0 c
c0
Inserting (10) into (14) and using the result from [8]
Z1
J n ðvÞ ¼ ln ttn1 evt dt; v [ 0; n ¼ 1; 2. . .
1
( ) ð16Þ
ðn 1Þ! X
n1
P k ðvÞ
¼ E1 ðvÞ þ
vn k¼1
k
R1
where E1(.) is the first order exponential integral function E1 ðxÞ ¼ 1 ext =t dt and P k ð:Þ is
P xj
the Poisson sum P k ðxÞ ¼ ex k1 j¼0 j! [14].
The capacity Copra can be written as
X1 kþ2mNr
Nt m 1 ðk þ 2mNr 1Þ!
Copra ¼ Nt c k kþ2mNr
CðmNr Þ k¼0 c c0 mNt
c
2 3 ð17Þ
kþ2mN
X r 1 P j
mNt c0
mNt c0 c
4E1 þ 5
c j¼1
j
To attain an optimal cutoff level c0 in (17), we require calculating for c0 in (15). Now,
R1
inserting (10) into (15) and using the result from [14] Cðn; xÞ ¼ sn1 es ds, the c0 is
x
obtained as
P1 kþ2mNr mN c
Nt c
k¼0 k
ðmcÞ Cfkþ2mNr ; ct 0 g
ÞNt
CðmNr ð c Þ
mNt
c0 ¼ P1 kþ2mNr n o ð18Þ
Nt ck ðmcÞ
1þ k¼0
CðmNr ÞNt
C k þ 2mNr 1; mNct c0
Thus, we notice that there is a unique c0 which satisfies (18). An asymptotic evaluation
of (18) shows that as c ! 1; c0 ! 1.
123
2238 S. S. Chauhan, S. Kumar
In this scheme, the transmitter utilizes the channel statistics fed back by the receiver and
modify its power to keep constant SNR at the receiver in order to invert the channel fading.
The average channel capacity Ccifr with this technique is given in [9–13] as
!
1
Ccifr ¼ log2 1 þ R 1 1 ð19Þ
0 c fcNt ðcÞdc
Inserting (10) into (19) and utilizing the integral from [14]
Z1
m bxn C mþ1n
x e dx ¼ ð20Þ
nbð n Þ
mþ1
The spectral efficiency obtained under cifr technique may further be raised by inverting
the channel only when the deep fade is above a cutoff level c0. This is referred as the
truncated inversion with fixed rate (tifr) technique [9–13]. The average channel capacity
Ctifr with this technique is given by
!
1
Ctifr ¼ log2 1 þ R 1 1 ð1 Pout Þ ð22Þ
c0 c fcNt ðcÞdc
Now, inserting (10) into (24) and using [14, (3.381, 3)],
Z1
xj1 eax dx ¼ aj Cðj; avÞ ð25Þ
v
"
1 X
1
mNt
Pout ¼ 1 ck C k þ 2mNr ; y ð26Þ
CðmNr ÞNt k¼0 c
123
Adaptive-Transmission Channel Capacity of Maximal-Ratio… 2239
Thus, inserting (26) into (22) and using (25), finally the channel capacity Ctifr can be
evaluated as
2 3
1
Ctifr ¼ log2 41 þ P1 5
c ð2k2mNr Þ mNt
CðmNr ÞNt
c N
k¼0 k t C k þ 2mNr 1; c
c 0
" # ð27Þ
X 1
1 mNt
ck C k þ 2mNr ; y
CðmNr ÞNt k¼0 c
4 Numerical Results
This section presents numerical results to illustrate and validate the exact closed-form
average channel capacity of TAS/MRC with AT techniques over Nakagami-m fading
channels.
The average channel capacity per unit bandwidth expression corresponding to Cora
technique as achieved from (13) is illustrated in Fig. 1 for (Nt, 1; Nr) TAS/MRC over
Nakagami-m fading channels with Nr ¼ 1; 2 Nt 3 for different fading parameters,
namely, m = 1 and m = 2. It can be observed that with increasing average received SNR
per channel and the number of transmit antennas, the spectral efficiency increases; how-
ever, it deteriorates as the fading severity increases.
Figures 2 and 3 present the comparison of the plots for ð2; 1; 1Þ and ð2; 1; 2Þ TAS/MRC
schemes in different AT techniques with fading parameter m = 0.8 and m = 1.6 respec-
tively. It can be illustrated that the exceptional capacity can be achieved under opra
7
Average Channel Capacity per unit Bandwidth [bit/sec/Hz]
m=2
5
2
m=1 Nt=2
Nt=3
1
0
0 2 4 6 8 10 12 14 16 18 20
Average Received SNR per Branch[dB]
Fig. 1 Average capacity per unit bandwidth for a TAS/MRC (Nt, 1; Nr) over Nakagami-
mðm ¼ 1 and m ¼ 2Þ fading channels with Nr = 1, 2 B Nt B 3
123
2240 S. S. Chauhan, S. Kumar
5
(2,1:1)
3
cifr
tifr
ora
2
opra
0
0 2 4 6 8 10 12 14 16 18 20
Average Received SNR per Branch[dB]
Fig. 2 Average capacity per unit bandwidth for a TAS/MRC (Nt, 1; Nr) over Nakagami-mðm ¼ 0:8Þ fading
channels with Nr ¼ 1 and Nt ¼ 2
scheme for any average received SNR per branch, as expected, since both the transmit
power and data rates adapts at the source according to the channel fading statistics. It can
be noticed that for high c values, the ora technique (13) realizes almost the same spectral
efficiency as the opra technique (17) when none power allocation at the transmitter side. It
can also be remarked that for any number of antennas and/or fading parameters, cifr
technique (21) experience the utmost capacity lost compare to the other techniques as large
amount of transmit power is required to compensate the fade depth. However, it is simpler
technique to implement, as it uses fixed coding and modulation. The spectral efficiency
achieved under cifr scheme can be further increases by using truncated version of cifr,
called tifr scheme whereby a cutoff level is used to maximize the spectral efficiency at the
penalty of an increased outage probability.
In Fig. 4, we compare the outage probabilities achieved under opra and tifr adaptive-
transmission schemes with Nr = 2 and various numbers of transmit antennas
½ðaÞNt ¼ 1; ðbÞNt ¼ 2 and ðcÞNt ¼ 4. This illustrates that the decrease in outage proba-
bility with increase in average received SNR per branch is slower for tifr than for opra. The
opra sustain an outage probability, which is always smaller than the outage probability
deteriorated by the tifr scheme for the same number of antenna since data transmission is
not valid when the c falls below the particular cutoff value.
5 Conclusions
123
Adaptive-Transmission Channel Capacity of Maximal-Ratio… 2241
(2,1:2)
5
3
cifr
tifr
2 ora
opra
0
0 2 4 6 8 10 12 14 16 18 20
Average Received SNR per Branch[dB]
Fig. 3 Average capacity per unit bandwidth for a TAS/MRC (Nt, 1; Nr) over Nakagami-mðm ¼ 1:6Þ fading
channels with Nr ¼ 2 and Nt ¼ 2
0
10
tifr
(a)
opra
10
-1 (b)
-2 (c)
10
Probability of outage
10
-3 (a)
-4
10
-5
10
-6 (b)
10
-7 (c)
10
-8
10
0 2 4 6 8 10 12 14 16 18 20
Average recieved SNR per branch (dB)
Fig. 4 Outage probability for opra and tifr scheme verses average received SNR per branch with Nr = 2
for a Nt = 1, b Nt ¼ 2 and c Nt = 4
123
2242 S. S. Chauhan, S. Kumar
technique and the cifr technique. Our outcomes depict that opra enrich a slight capacity
growth, as both transmit power and rate adapts at the source and ora technique obtained
almost the same capacity as the opra policy, mainly for high c values. For any number of
antennas and/or fading parameters cifr technique has lowest capacity and can further be
raised by using a tifr technique where a particular cutoff level is chosen to enhance the
channel capacity. Hence, it can be concluded that by increasing the average received SNR
per branch, fading parameter m and the diversity gain, the channel capacity of the system
increases.
References
1. Telatar, I. E. (1999). Capacity of multi-antenna Gaussian channels. Eur. Trans. Telecommun. Relat.
Technol., 10, 585–595.
2. Foschini, G. J., & Gans, M. J. (1998). On limits of wireless communication in a fading environment
when using multiple antennas. Springer Wireless Pers. Commun., 6, 311–355.
3. Molisch, A. F., Win, M. Z., Choi, Y. S., & Winters, J. H. (2005). Capacity of MIMO systems with
antenna selection. IEEE Transactions on Wireless Communications, 4(4), 1759–1762.
4. Yang, L., & Qin, J. (2006). Performance of the Alamouti scheme with transmit antenna selection for
M-ary Signals. IEEE Transactions on Wireless Communications, 5(12), 3365–3369.
5. Chen, Z., Chi, Z., Li, Y., & Vucetic, B. (2009). Error performance of maximal-ratio combining with
transmit antenna selection in flat Nakagami-m fading channels. IEEE Transactions on Wireless Com-
munications, 8(1), 424–431.
6. Romero-Jerez, J. M., & Goldsmith, A. J. (2009). Performance of multichannel reception with transmit
antenna selection in arbitrarily distributed Nakagami fading channels. IEEE Transactions on Wireless
Communications, 8(4), 2006–2013.
7. Tan, B. S., Li, K. H., & Teh, K. C. (2012). Performance study of transmit antenna selection with switch-
and-examine combining over Rayleigh fading. IEEE Trans. Veh. Tech., 61(9), 4205–4211.
8. Alouini, M. S., & Goldsmith, A. J. (1999). Capacity of Rayleigh fading channels under different
adaptive transmission and diversity-combining techniques. IEEE Transactions on Vehicular Technol-
ogy, 48(4), 1165–1181.
9. Maaref, A., & Aı̈ssa, S. (2005). Capacity of space time block codes in MIMO Rayleigh fading with
adaptive transmission and estimation errors. IEEE Transactions on Wireless Communications, 4(5),
2568–2578.
10. Pan, G., Ekici, E., & Feng, Q. (2012). Capacity analysis of log-normal channels under various adaptive
transmission schemes. IEEE Communications Letters, 16(3), 346–348.
11. Zhang, Y., & Tepedelenioglu, C. (2012). Asymptotic capacity analysis for adaptive transmission
schemes under general fading distributions. IEEE Transactions on Information Theory, 58(2), 897–908.
12. Chauhan, S. S., & Kumar, S. (2014). Capacity analysis of adaptive transmission with space-time block
codes in spatially correlated MIMO Nakagami-m fading channels. Springer Wireless Pers. Commun.,
79(2), 1211–1222.
13. Chauhan, S. S. & Kumar, S. (2015). Capacity of orthogonal space-time block codes in spatially
correlated MIMO Weibull fading channel under various adaptive transmission techniques. Accepted in
Springer. Tele. Syst.
14. Gradshteyn, I. S., & Ryzhik, I. M. (2000). Table of Integrals, Series, and Products (6th ed.). San Diego:
Academic.
15. Simon, M. K., & Alouini, M. S. (2000). Digital Communication over Fading Channels: A Unified
Approach to Performance Analysis. New York: Wiley.
123
Adaptive-Transmission Channel Capacity of Maximal-Ratio… 2243
123