Conference PaperPDF Available

Energy Efficiency Analysis of Rank-1 Ricean Fading MIMO Channels

Authors:
  • Ericsson AB, Gothenburg, Sweden

Abstract and Figures

This paper studies the energy efficiency (EE) of a point-to-point rank-1 Ricean fading multiple-input-multiple-output (MIMO) channel. In particular, a tight lower bound and an asymptotic approximation for the EE of the considered MIMO system are presented, under the assumption that the channel is unknown at the transmitter and perfectly known at the receiver. Moreover, the effects of different system parameters, namely, transmit power, spectral efficiency (SE), and number of transmit and receive antennas, on the EE are analytically investigated. An important observation is that, in the high signal-to-noise ratio regime and with the other system parameters fixed, the optimal transmit power that maximizes the EE increases as the Ricean-K factor increases. On the contrary, the optimal SE and the optimal number of transmit antennas decrease as K increases.
Content may be subject to copyright.
Chalmers Publication Library
Copyright Notice
©2014 IEEE. Personal use of this material is permitted. However, permission to
reprint/republish this material for advertising or promotional purposes or for creating new
collective works for resale or redistribution to servers or lists, or to reuse any copyrighted
component of this work in other works must be obtained from the IEEE.
This document was downloaded from Chalmers Publication Library (http://publications.lib.chalmers.se/), where it
is available in accordance with the IEEE PSPB Operations Manual, amended 19 Nov. 2010, Sec. 8.1.9
(http://www.ieee.org/documents/opsmanual.pdf)
(Article begins on next page)
Energy Efficiency Analysis of Rank-1
Ricean Fading MIMO Channels
Jingya Li, Michail Matthaiou?, Shi Jin, and Tommy Svensson
Department of Signals and Systems, Chalmers University of Technology, Gothenburg, Sweden
?School of Electronics, Electrical Engineering and Computer Science, Queen’s University Belfast, Belfast, U.K.
National Mobile Communications Research Laboratory, Southeast University, Nanjing, China
Email: jingya.li@chalmers.se, m.matthaiou@qub.ac.uk, jinshi@seu.edu.cn, tommy.svensson@chalmers.se
Abstract—This paper studies the energy efficiency (EE) of
a point-to-point rank-1 Ricean fading multiple-input-multiple-
output (MIMO) channel. In particular, a tight lower bound and
an asymptotic approximation for the EE of the considered MIMO
system are presented, under the assumption that the channel is
unknown at the transmitter and perfectly known at the receiver.
Moreover, the effects of different system parameters, namely,
transmit power, spectral efficiency (SE), and number of transmit
and receive antennas, on the EE are analytically investigated. An
important observation is that, in the high signal-to-noise ratio
regime and with the other system parameters fixed, the optimal
transmit power that maximizes the EE increases as the Ricean-K
factor increases. On the contrary, the optimal SE and the optimal
number of transmit antennas decrease as Kincreases.
I. INTRODUCTION
MIMO techniques can greatly improve the SE of wireless
communications systems. However, due to the extra signal
processing and circuit power consumption per transceiver RF
chain, the use of multiple antennas also leads to an increased
energy consumption. In future green communication systems,
the EE becomes an important performance metric. Therefore,
the trade-off between EE and SE has become a hot topic in
MIMO systems.
Most studies on the EE of MIMO systems adopted the
assumption of Rayleigh fading conditions [1]–[3]. However,
in many practical scenarios, there exists a deterministic or
strong line-of-sight (LoS) component. Some relevant examples
are: 1) short-range millimeter wave MIMO channels [4],
2) microwave backhauling between macro base stations and
outdoor small-cell base stations [5], 3) the channel between a
user and its connected small-cell base stations. In such cases,
Ricean fading conditions should be considered, which induces
several research challenges.
Motivated by the above discussion, in this paper, we study a
worst case scenario, where the Ricean fading MIMO channel
has a rank-1 mean. This scenario is relevant when the inter-
element distances at both ends are small and the impinging
LoS waveforms have identical phases [6]. A tight lower
bound for and an asymptotic approximation on the EE of the
considered MIMO system are presented, assuming that the
This work was supported in part by the Swedish Governmental Agency
for Innovation Systems (VINNOVA) within the VINN Excellence Center
Chase, and the Swedish Research Council VR under the project 621-2009-
4555 Dynamic Multipoint Wireless Transmission.
channel is unknown at the transmitter and perfectly known at
the receiver. With the help of the closed-form EE expressions,
the effects of different system parameters, i.e., transmit power,
SE, and number of transmit and receive antennas, on the EE
are analytically investigated. Interestingly, we observe that,
in the high signal-to-noise ratio (SNR) regime and with the
other system parameters fixed, the optimal transmit power that
maximizes the EE increases with the Ricean-Kfactor or with
the fixed transceiver power consumption. On the other hand,
the optimal SE and the optimal number of transmit antennas
decrease as the Ricean-Kfactor increases, i.e., when the LoS
component becomes more and more dominant.
II. SY STE M MOD EL
We consider a point-to-point MIMO system with nttransmit
antennas and nrreceive antennas. Here, we assume ntnr,
while all our results can be easily generalized to the case of
ntnr. Let sCnt×1denote the transmitted signal vector
with EssP, where (·)stands for conjugate transpose.
The received signal vector at the receiver, yCnr×1, is
y=Hs +n(1)
where nCnr×1denotes the circularly symmetric complex
Gaussian noise with n CN 0, σ2Inr. The MIMO channel,
HCnr×nt, is assumed to be Ricean flat-fading with rank-1
mean, such that
H=rK
K+ 1 ¯
H+r1
K+ 1 ˜
H(2)
where E{H}=qK
K+1 ¯
Hand Kis the Ricean K-factor.
Here, E{·} stands for expectation. The rank-1 matrix ¯
H
corresponds to the coherent line-of-sight component, and it
is normalized such that
¯
H
2
F=ntnr, where k·kFde-
notes the Frobenius norm. The term ˜
Hrepresents the non-
coherent scattered contributions. In this paper, we focus on
the case without spatial correlation between receive antennas
and transmit antennas. Thus, ˜
Hcan be modeled as a random
fading matrix, whose entries are circularly symmetric complex
Gaussian random variables with unit variance. Therefore,
Hhas a matrix-variate complex Gaussian distribution with
H CN M,Inrε2Int, where ε,1/K+ 1 and
M,εK¯
H. Here, is defined as the Kronecker product
between matrices.
Under the assumption of no channel state information (CSI)
at the transmitter and perfect CSI at the receiver, the SE, which
is defined as the ergodic capacity per unit bandwidth, is [7]
SE =EHlog2det Int+P
σ2nt
HH (3)
where the expectation is taken with respect to Hand det (·)
denotes the matrix determinant. The EE is defined as the ratio
between the SE and total power consumption, that is,
EE =SE
Ptot
.(4)
In this paper, we adopt the linear approximated power con-
sumption model proposed in [8], where
Ptot =αP +ntPct +nrPcr.(5)
The static terms, Pct and Pcr, denote the fixed power con-
sumption for each transmit and receiver chain, respectively.
The scaling factor, α, models the impact of the output power
Pon the efficiency of the power amplifiers used at the
transmitter. The values of Pct,Pcr and αfor different base
station (BS) types can be found in [8, Table 2].
III. SPECTRAL RES ULTS
The exact expressions for SE and EE are difficult to obtain.
To make the EE analysis tractable, we consider the following
two approximations of the SE.
Lemma 1 [9, Theorem 9]: The SE of the uncorrelated Ricean
fading MIMO channel with rank-1 mean matrix Mand nt
nr, is tightly lower bounded for arbitrary SNR =P
σ2by SE
SE1, where
SE1,
nt
X
i=1
log2(1 + µiP)(6)
with positive values of µigiven as
µi=
exp(gnr(4))
σ2nt(K+1) , i = 1
exp(Pnri
p=1
1
pγ)
σ2nt(K+1) , i = 2, . . . , nt
where γ= 0.5772... is Euler’s constant and 4,Kntnr. The
function gnr(4)is defined as
gnr(4),ln (4)Ei (−4) +
nr1
X
i=1 1
4i
×e−4 (i1)! (nr1)!
i(nr1i)!(7)
where Ei (x) = ´
x
et
tdt,x > 0, is the exponential
integral function.
Note that for the special case of K , (6) reduces to
SEK→∞
1= log21 + nr
P
σ2(8)
which is the exact capacity of deterministic MIMO systems,
with multiplexing gain equal to 1 [7]. In the high SNR regime,
we have the following approximation for the SE.
Lemma 2 [9, Eq. (24)]: The SE of the uncorrelated Ricean
fading MIMO channel in the high SNR regime, with rank-1
mean matrix Mand ntnr, can be approximated as SE
SE2, where
SE2,ntlog2P
σ2nt+c(9)
where c,log2nr!
(nrnt)! + log21+ntK
(K+1)nt.
IV. ENERGY EFFIC IEN CY MA XIM IZ ATION
In this section, we utilize the closed-form expressions in
Section III to analyze the impact of different system parame-
ters on the EE of rank-1 Ricean fading MIMO channels.
A. The Impact of Transmit Power on EE
1) Arbitrary SNR: We start with finding the optimal trans-
mit power Pby using the lower bound EE expression, i.e.,
EE1,SE1
Ptot
=Pnt
i=1 log2(1 + µiP)
αP +ntPct +nrPcr
(10)
which is tight for arbitrary SNR. From (10), we see that the
sublevel sets Sν={−EE1ν}={Pnt
i=1 log2(1 + µiP)
+ν(αP +ntPct +nrPcr)0}are convex for νR. Thus,
EE1is a quasi-concave function of P. The first derivative of
EE1with respect to Pis
EE1
∂P =f1(P)
(αP +ntPct +nrPcr)2ln 2 (11)
where
f1(P),(αP +ntPct +nrPcr)
nt
X
i=1
µi
1 + µiP
α
nt
X
i=1
ln (1 + µiP)(12)
which is a decreasing function of P. Note that f1(0) =
(ntPct +nrPcr)Pnt
i=1 µi>0. Thus, there exists a unique
Psuch that f1(P) = 0, i.e., EE1
∂P |P=P= 0. Recall that
EE1in (10) is a quasi-concave function of P. Thus, the unique
Pis the global optimal power which maximizes EE1. Unfor-
tunately, the optimal value of P= arg (f1(P) = 0) cannot
be expressed in closed-form. However, with the expression in
(12) and noting that f1(P)is a decreasing function of P, we
can easily obtain the optimal power value numerically, e.g.,
via a bisection search shown in Algorithm 1.
Algorithm 1 Bisection method for finding the optimal transmit
power that maximizes EE1
given PlPand PuP, with f1(Pl)0and f1(Pu)0,
given tolerance > 0.
repeat
1: set P=Pl+Pu
2, and calculate f1(P)by using (12).
2: if f1(P)>0then
3: Pl:= P
4: else
5: Pu:= P
until |f1(P)|< .
An initial lower bound can be easily found as Pl= 0.
To find an initial upper bound Pusuch that g(Pu)0, we
define µmax ,max (µi)and µmin ,min (µi). Then, we have
f1(P)nt˜
f1(P), where
˜
f1(P),(αP +ntPct +nrPcr)µmax
1 + µminP
αln (1 + µminP).(13)
Therefore, an upper bound Pucan be found by setting
˜
f1(Pu) = 0, from which we get
Pu=
exp x+µmax
µmin 1
µmin
(14)
where
x,W
µmax (αP +ntPct +nrPcr)
αexp µmax
µmin
(15)
and Wis the Lambert Wfunction satisfying W(x)eW(x)=x.
With the closed-form expression of Puin (14), Algorithm 1
is guaranteed to converge to the optimal P.
2) High SNR Approximation: In order to gain more physi-
cal insights, we now elaborate on the high SNR regime. Plug-
ging (9) into (4), the EE can be approximated as EE EE2,
where
EE2,SE2
Ptot
=
ntlog2P
ntσ2+c
αP +ntPct +nrPcr
.(16)
Proposition 1: For the uncorrelated Ricean fading MIMO
channel with rank-1 mean matrix Mand nrnt, in the high
SNR regime, the optimal transmit power that maximizes EE2
is
P=ntσ2exp WntPct +nrPcr
ntσ2αed+d (17)
where d,1c
ntln 2, and cis defined in Lemma 2.
Proof: From (16), we see that EE2is a quasi-concave
function of P. By calculating the first order derivative of EE2
with respect to P, and setting it to zero, we get
f2(P) = ntPct +nrPcr
αP ln P
ntσ2d= 0.(18)
Let z= ln P
ntσ2d, then (18) can be rewritten as zez=
ntPct+nrPcr
ntσ2αed, from which, we get z=WntPct+nrPcr
ntσ2αed.
Therefore, the optimal transmit power can be obtained by
substituting the value of zinto P=ntσ2exp (z+d). The
optimal solution (17) can also be found by using a different
methodology presented in [10].
Note that W(x)is an increasing function of xfor x > 0.
Thus, from Proposition 1, we observe that in the high SNR
regime, the optimal transmit power increases with increasing
the fixed power consumption Pct and/or Pcr. Moreover, with
the aid of [1, Lemma 3], we can show that the optimal transmit
power is approximately equal to ntσ2edfor small ntPct +
nrPcr. For large ntPct +nrPcr, the optimal transmit power
increases linearly with ntPct+nrPcr . This is intuitive, because
when ntPct +nrPcr is small, the effect of the fixed power
consumption on the EE is negligible. On the contrary, when
the fixed power consumption is large and dominates the total
power consumption, then more transmit power can be used to
increase the SE, thereby, increasing the EE.
Corollary 1: The optimal transmit power, Pin (17),
increases as the Ricean K-factor, K, increases.
Proof: Note that c, which is defined in Lemma 2, de-
creases with increasing K. Thus, d,1c
ntln 2 increases
as Kincreases. Then, from (18), we see that f2(P)is an
increasing function of K. Note that f2(P)is a decreasing
function of P. Therefore, the optimal transmit power, P,
which satisfies f2(P)=0, increases as Kincreases.
Corollary 1 implies that, in the high SNR regime, in order
to maximize the EE, the transmit power should increase in
order to compensate for the SE loss due to the dominating
LoS component.
B. The Impact of SE on EE
For arbitrary SNR, the optimal SE that maximizes EE1can
be obtained by substituting Pobtained from Algorithm 1
into (6). For the high SNR regime, we present the following
proposition.
Proposition 2: For the uncorrelated Ricean fading MIMO
channel with rank-1 mean matrix Mand nrnt, in the high
SNR regime, the optimal value of SE2that maximizes EE2is
SE
2=nt
ln 2 W(ntPct +nrPcr)ed
ntασ2+ 1.(19)
Proposition 2 implies that increasing the SE does not always
increase the EE. Recall that in the high SNR regime, the
optimal transmit power increases as ntPct +nrPcr increases.
As expected, the optimal SE also increases with increasing
the fixed power consumption. For small ntPct +nrPcr, the
optimal SE is approximately equal to nt
ln 2 , which increases
linearly with nt. However, when the fixed power consumption
is large, the optimal SE increases only logarithmically with
ntPct +nrPcr. Recall that W(x)is an increasing function of
xfor x > 0, and dis an increasing function of K. From (19),
we have the following corollary.
Corollary 2: The optimal SE, SE
2in (19), decreases as the
Ricean K-factor, K, increases.
Corollary 2 claims that, in the high SNR regime, the SE
decreases as the LoS component becomes more and more
dominant, which is expected due to the reduced multiplexing
gain [9].
C. The Impact of Number of Antennas on EE
Since the optimal ntand nrthat maximize EE1for arbitrary
SNR are difficult, if not impossible, to obtain, we now focus
on the high SNR regime to gain insights. The values of nt
and nrare assumed to be continuous with 0< ntnr.
1) The Impact of nton EE: We start with investigating
the impact of the number of transmission antennas, nt, on the
EE. The number of receive antennas, nr, is assumed to be
fixed. Utilizing Stirling’s approximation, ln n!nln nn,
(9) becomes
SE21
ln 2 ntln P
σ2nt(1 + K)1+ ln (1 + ntK)
+1
ln 2 nrln nr(nrnt) ln (nrnt).(20)
By doing so, it can be shown that EE2in (16) is a quasi-
concave function of nt, with
EE2
∂nt(αP +nrPcr )
(αP +ntPct +nrPcr)2ln 2 J1(nt)(21)
where
J1(nt),(1 + nrβ) ln (nrnt)ln ntβln (1 + ntK)
+(1 + ntβ)K
1 + ntK+ ln P
σ2(1 + K)1βnrln nr
(22)
with β,Pct
αP +nrPcr . Note that J1(nt)is a decreasing function
of nt. Recall that 0< ntnrand note that J1(nt0) >0
and J1(ntnr)<0. Thus, there exists a unique optimal ˆnt,
where
ˆnt= arg
0<ntnr
(J1(nt) = 0) .(23)
The optimal value of ˆntcannot be expressed in closed-form.
However, similar to Algorithm 1, we can easily obtain ˆnt
numerically via a bisection search. If the value of ˆntis non-
integer, the optimal n
tis attained at one of the two closest pos-
itive integers. Note that J1(nt)is also a decreasing function
of βand K. Therefore, ˆntdecreases as βor/and Kincreases.
Intuitively, this implies that, in the high SNR regime, the
optimal number of transmit antennas decreases when the fixed
power consumption at the transmitter increases. Moreover, it
indicates that less transmit antennas should be used when the
LoS component becomes more and more dominant. This is
expected since when the Ricean K-factor is large, the SE gain
achieved by adding more transmit antennas is negligible. Note
that the total power consumption increases with increasing nt.
Thus, as Kincreases, less transmit antennas should be used in
order to achieve a better trade-off between the SE and power
consumption.
2) The Impact of nron EE: Now, we try to find the optimal
number of receive antennas, nr, by keeping ntfixed. Utilizing
(20), it can been shown that EE2in (16) is also a quasi-concave
function of nr, with
EE2
∂nrαP +ntPct
(αP +ntPct +nrPcr)2ln 2 J2(nr)(24)
where
J2(nr),ln nra1ln (nrnt)a2(25)
with a1,1 + ntζand a2,
ζntln P
ntσ21+ ln 1+ntK
(1+K)nt >0in the high
SNR regime. Here, ζ,Pcr
αP +ntPct . It is easy to show that
J2(nr)is a decreasing function of nr. Recall that nrnt
and noting that J2(nr+)<0and J2(nrnt)>0.
Thus, there exists a unique optimal ˆnr, where
ˆnr= arg
nrnt
(J2(nr) = 0) .(26)
The optimal value of ˆnrcannot be expressed in closed-
form. However, similar to Algorithm 1, we can easily obtain
ˆnrnumerically via a bisection search. Note that J2(nr)
ln nr
nrnta2. Thus, by setting ln nr
nrnta2= 0, an upper
bound of ˆnrcan be found as ˆnrnt
1ea2. Finally, the optimal
n
ris attained at one of the two closest integers.
From (25), we see that J2(nr)decreases as ζincreases.
Therefore, in order to satisfy J2nr)=0,ˆnrshould decrease
as ζincreases. This implies that the optimal number of
receive antennas decreases with increasing the fixed power
consumption at the receiver. Note that a2is proportional to
ζ. Thus, when the fixed power consumption at the receiver
side dominates the total power consumption, the upper bound,
nt
1ea2will approach to nt. In this case, we have n
r=nt.
3) Special Case of nt=nr:For the special case of nt=
nr=n, we present the following corollary.
Proposition 3: When nt=nr=n, the optimal nthat
maximizes EE2is given by n= arg max
{bˆnc,dˆne}
EE2, where
ˆn=
exp W αP K
Pct+Pcr 11
eb+b1
K(27)
with b,1 + αP
Pct+Pcr ln P
σ2(1+K)1.
Proof: Plugging nt=nr=ninto (9),
and utilizing Stirling’s approximation, we get SE2
1
ln 2 nln ρ
1+Kn+ ln (1 + nK), which is a concave func-
tion of n. Therefore, EE2in (16) is a quasi-concave function
of n, with EE2
∂n J3(n)
(αP +n(Pct+Pcr ))2ln 2 , where
J3(n),αP ln P
σ2(1 + K)1 + K
1 + nK
(Pct +Pcr)ln (1 + nK)nK
1 + nK .(28)
It can be shown that J3(n)is a decreasing function of n, with
J3(n= +)<0and J3(n= 1) >0. Thus, there exists a
unique optimal ˆnsuch that J3n) = 0, and the optimal nis
attained at one of the two closest integers from ˆn.
Similarly, utilizing (28), we can show that the optimal
number of antennas, ˆn, decreases as the Ricean K-factor
and/or the fixed power consumption Pct +Pcr increase.
V. NUMERICAL RE SU LTS
Numerical results are presented to verify our analytical re-
sults and to illustrate the impact of different system parameters
(P,SE,ntand nr) on the EE of rank-1 Ricean fading MIMO
systems. The noise power is σ2= 10mW.
Figure 1 shows the EE as a function of the transmit power,
P, for different Ricean-Kfactors. The numbers of transmit
and receive antennas are nt= 2 and nr= 4, respectively. The
power model parameters are Pct = 56W, Pcr = 130W and
α= 2.6, which correspond to a backhaul channel between a
micro BS to a macro BS [8, Table 2]. The lower bound EE1and
the high SNR approximated EE2are compared with Monte-
Carlo simulations. We see that both analytical EE expressions
agree perfectly with the numerical results. As expected, the
EE decreases as the Ricean-Kfactor increases. Utilizing
Proposition 1, for K= 0,10,100, the optimal power values
are 32.45, 35.49 and 41.14W, respectively, which agree with
the simulated optimal values as shown in Fig. 1. Moreover,
in agreement with Corollary 1, we observe that the optimal
transmit power increases as the Ricean-Kfactor increases.
This indicates that, in order to maximize the EE, the transmit
power should increase so as to compensate for the SE loss
due to the increased LoS component.
Figure 2 demonstrates the EE versus the SE considering the
same system model parameters used in Fig. 1. We see that even
though the optimal transmit power increases as Kincreases,
the optimal value of SE decreases when increasing K. This
observation is in agreement with Corollary 2. Moreover, we
see that when the SE is less than 12 bits/s/Hz, the EE increases
linearly with the SE. This is because, for small values of
SE, the corresponding transmit power Pis small. Thus, by
0 10 20 30 40 50 60
0
0.005
0.01
0.015
0.02
0.025
0.03
0.035
0.04
X: 31.98
Y: 0.03416
Transmit power, P[W]
EE [bits/Hz/Joule]
X: 35.88
Y: 0.0309
X: 41.14
Y: 0.02667
Simulation
EE1
EE2
K=0
K=10
K=10 0
Fig. 1. Energy Efficiency vs. transmit power.
0 4 8 12 16 20 24 28
0
0.005
0.01
0.015
0.02
0.025
0.03
0.035
SE [bits/s/Hz]
EE [bits/Hz/Joule]
Simulation
EE1
EE2
K=10 0
K=0
K=10
Fig. 2. Energy efficiency vs. spectral efficiency.
increasing P, the increase of the total power consumption Ptot
is negligible. Therefore, the EE, which is defined as the SE-
to-Ptot ratio, increases linearly as the SE increases.
Finally, Fig. 3 investigates the impact of the number of
transmit antennas on the EE of rank-1 Ricean fading MIMO
channels. The number of receive antennas nris set to 10.
The EE is plotted as a function of ntfor different values of
Kand Pcr respectively. We consider a micro BS transmitter
(Pct = 56W and α= 2.6), and investigate three different
backhaul MIMO channels with Pcr = 6.8,56 and 130W,
which correspond to three different types of receivers, i.e.,
pico BS, micro BS and macro BS, respectively [8, Table 2].
As expected, for a fixed value of Pcr, the optimal ntdecreases
as Kincreases, since with increasing the Ricean K-factor,
the contribution to the SE by adding more transmit antennas
becomes negligible, while the total power consumption will
increase linearly. On the other hand, for a fixed value of K,
the optimal ntincreases as Pcr increases. This observation
is also intuitive, because with increasing Pcr, the total power
consumption becomes dominated by the power consumption at
the receiver. Therefore, more transmit antennas can be added
to increase the SE, thereby, increasing the EE.
1 2 3 4 5 6 7 8 9 10
0.06
0.07
0.08
0.09
0.1
0.11
0.12
0.13
0.14
0.15
Number of transmit antennas, nt
EE [bits/Hz/Joule]
Simulation
EE1
EE2
12345678910
0
0.05
0.1
0.15
Number of transmit antennas, nt
EE [bits/Hz/Joule]
Simulation
EE1
EE2
K=0, Pcr=56W
K=10 , Pcr=56W
K=10 0, Pcr=56 W
K=10 , Pcr=6.8W
K=10 , Pcr=56W
K=10 , Pcr=130 W
Fig. 3. Energy efficiency vs. the number of transmit antennas.
VI. CONCLUSIONS
In this paper, the EE of rank-1 Ricean fading MIMO chan-
nels was analyzed. More specifically, the effects of different
system parameters, i.e., transmit power, SE, and number of
transmit and receive antennas, on the EE were investigated.
When keeping the other parameters fixed, closed-from expres-
sions have been derived for the optimal transmit power and
the optimal SE, respectively, with the objective of maximizing
the EE in the high SNR regime. We analytically show that
the optimal transmit power and the optimal SE increase as
the fixed power consumption per transceiver chain increases.
Moreover, both our theoretical analysis and numerical results
indicated that, as the Ricean-Kfactor increases, the optimal
transmit power will increase; on the contrary, the optimal SE
and the optimal number of transmit antennas will decrease
with increasing K.
REFERENCES
[1] E. Björnson, L. Sanguinetti, J. Hoydis, and M. Debbah, “Designing
multi-user MIMO for energy efficiency: When is massive MIMO the
answer?” in Proc. IEEE Wireless Commun. Netw. Conf. (WCNC), Apr.
2014. Available: http://arxiv.org/abs/1310.3843.
[2] J. Jiang et al., “On the relation between energy efficiency and spectral
efficiency of multiple-antenna systems, IEEE Trans. Veh. Technol.,
vol. 62, no. 7, pp. 3463–3469, Sept. 2013.
[3] F. Heliot, M. Imran, and R. Tafazolli, “On the energy efficiency-spectral
efficiency trade-off over the MIMO Rayleigh fading channel,” IEEE
Trans. Commun., vol. 60, no. 5, pp. 1345–1356, May 2012.
[4] E. Torkildson, U. Madhow, and M. Rodwell, “Indoor millimeter wave
MIMO: Feasibility and performance,” IEEE Trans. Wireless Commun.,
vol. 10, no. 12, pp. 4150–4160, Dec. 2011.
[5] Ericsson, “It all comes back to backhaul,” White Paper, Feb.
2012. Available: http://www.ericsson.com/res/docs/whitepapers/WP-
Heterogeneous-Networks-Backhaul.pdf.
[6] I. Sarris, A. R. Nix, and A. Doufexi, “High-throughput multiple-input
multiple-output systems for in-home multimedia streaming,” IEEE
Wireless Commun., vol. 13, no. 5, pp. 60–66, Oct. 2006.
[7] I. E. Telatar, “Capacity of multi-antenna Gaussian channels, Europ.
Trans. Telecommun., vol. 10, no. 6, pp. 585–595, Nov.-Dec. 1999.
[8] G. Auer et al., “How much energy is needed to run a wireless network?”
IEEE Wireless Commun., vol. 18, no. 5, pp. 40–49, Oct. 2011.
[9] S. Jin, X. Gao, and X. You, “On the ergodic capacity of rank-1 Ricean-
fading MIMO channels,” IEEE Trans. Inf. Theory, vol. 53, no. 2, pp.
502–517, Feb. 2007.
[10] C. Isheden et al., “Framework for link-level energy efficiency optimiza-
tion with informed transmitter, IEEE Trans. Wireless Commun., vol. 11,
no. 8, pp. 2946–2957, Aug. 2012.
... Thus, we focus herein on evaluating MIMO spatial multiplexing with ZF under Rician fading that is transmit-correlated. For tractable analysis, as in [15] [ 16], we assume that the LoS or deterministic component of H satisfies rank(H d ) = r = 1. Whereas for LoS propagation r can take any value from 1 to N T [25], small antenna apertures, relatively-low carrier frequency, or large transmitter-receiver distance, as in point-to-point deployments [1] [16], are likely to yield H d as outer product of array response vectors [1,Eq. ...
... For tractable analysis, as in [15] [ 16], we assume that the LoS or deterministic component of H satisfies rank(H d ) = r = 1. Whereas for LoS propagation r can take any value from 1 to N T [25], small antenna apertures, relatively-low carrier frequency, or large transmitter-receiver distance, as in point-to-point deployments [1] [16], are likely to yield H d as outer product of array response vectors [1,Eq. (7.29), p. 299], which implies r = 1. ...
... 6 Without loss of generality, because of the full-Rician model adopted in (9). (16) and partition according to (12) the matrix ...
Article
Full-text available
We propose a new exact analysis and evaluation of zero-forcing detection (ZF) for multiple-input/multiple-output (MIMO) spatial multiplexing under transmit-correlated Rician fading for an NR×NT N_R \times N_T channel matrix with rank-1 line-of-sight (LoS) component. First, an analysis based on several matrix transformations yields the exact signal-to-noise ratio (SNR) moment generating function (m.g.f.) as an infinite series of gamma distribution m.g.f.'s. This produces analogous series for the SNR probability density function and for ZF performance measures. However, their numerical convergence is inherently problematic with increasing Rician K -factor, NR N_R , and NT N_T . Therefore, we additionally derive corresponding differential equations by using computer algebra. Finally, we apply the holonomic gradient method (HGM), i.e., we solve the differential equations by starting from suitable initial conditions computed with the infinite series. HGM yields more reliable performance evaluation than by infinite series alone and more expeditious than by simulation, for realistic values of K , and even for NR N_R and NT N_T relevant to large MIMO systems. We anticipate that future MIMO analyses for Rician fading will produce even more involved series that may benefit from the proposed computer-algebra-aided evaluation approach.
... The total power consumption of the network can be modeled with a circuit part that depends on the transceiver hardware and a dynamic part that is a function of the transmitted signal power [10]- [13]. Adding more low-power BSs can reduce the dynamic power consumption due to the shorter propagation distances, but require more hardware; thus, it will increase the circuit power part. ...
Article
Full-text available
This paper considers a downlink heterogeneous network, where different types of multi-antenna base stations (BSs) communicate with a number of single-antenna users. Multiple BSs can serve the users by spatial multiflow transmission techniques. Assuming imperfect channel state information at both BSs and users, the precoding, load balancing, and BS operation mode are jointly optimized for improving the network energy efficiency. We minimize the weighted total power consumption while satisfying quality of service constraints at the users. This problem is non-convex, but we prove that for each BS mode combination, the considered problem has a hidden convexity structure. Thus, the optimal solution is obtained by an exhaustive search over all possible BS mode combinations. Furthermore, by iterative convex approximations of the non-convex objective function, a heuristic algorithm is proposed to obtain a suboptimal solution of low complexity. We show that although multi-cell joint transmission is allowed, in most cases, it is optimal for each user to be served by a single BS. The optimal BS association condition is parameterized, which reveals how it is impacted by different system parameters. Simulation results indicate that putting a BS into sleep mode by proper load balancing is an important solution for energy savings.
Article
Purpose: The association of dairy products consumption with risk of obesity remains controversial. Therefore, we reviewed and quantitatively synthesized the evidence from observational studies with a meta-analysis. Methods: A literature search was performed in relevant databases. Random-effects model was used to pool odds ratios with 95% confidence intervals. Dose-response relationship was assessed by restricted cubic spline model. Results: Seventeen studies for total dairy products and 16 studies for milk with risk of obesity were eligible. The pooled odds ratios (95% confidence intervals) of obesity for the highest versus lowest category of total dairy products consumption were 0.54 (0.38-0.77) in children, 0.75 (0.69-0.81) in adults, and 0.74 (0.68-0.80) for both. Evidence of a nonlinear relationship was found (Pfor nonlinearity = .009). Milk consumption was also associated with risk of obesity [0.81 (0.75-0.88)] both in children [0.87 (0.80-0.95)] and in adults [0.77 (0.68-0.87)], and a linear relationship (Pfor nonlinearity = .598) suggested that risk of obesity decreased by 16% [0.84 (0.77-0.92)] for every 200 g/d increment of milk consumption. Conclusions: This meta-analysis indicates that dairy products consumption may be associated with a decreased risk of obesity. This association may be of public health significance.
Article
Aims To estimate the incidence of Type 2 diabetes mellitus (T2DM) and the number of those with T2DM attributable to overweight and obesity in China. Methods We conducted a prospective cohort study among 15 680 participants (46.4%, men) aged 35-74 years. The mean duration of follow-up was 8.0 years. We examined the relationship between overweight, obesity and risk of T2DM by Cox proportional hazards models. Population attributable risk (PAR) of overweight and obesity was also calculated. Moreover, we estimated the number of T2DM events attributed to overweight and obesity using PAR, incidence of T2DM and the population size of China in 2010. Results During a mean follow-up of 8.0 years, the age-standardized incidence of T2DM was 9.5 per 1 000 person-years in men and 9.2 in women. Overweight accounted for 28.3% (95% confidence interval [CI]: 20.1, 36.2) of incident T2DM among men and 31.3% (95% CI: 25.5, 36.9) among women. The corresponding PAR of obesity was 10.1% (95% CI: 6.0, 14.2) among men and 16.8% (95% CI: 12.0, 21.6) among women. Approximately 3.32 million (95% CI: 2.47, 4.24) incident T2DM were attributable to overweight and obesity in Chinese adults who were 35 to 74 years in 2010. Conclusion Our results indicate that incident T2DM is mainly attributable to overweight and obesity in China. It is extremely important to advocate healthy lifestyle and prevent excessive weight gain for reducing T2DM burden in China.
Article
Full-text available
Motivated by the increasing interest in energy-efficient communication systems, the relation between energy efficiency (EE) and spectral efficiency (SE) for multiple-input-multiple-output (MIMO) systems is investigated in this paper. To provide insights into the design of practical MIMO systems, we adopt a realistic power model and consider both independent Rayleigh fading and semicorrelated fading channels. We derived a novel and closed-form upper bound (UB) for the system EE as a function of SE. This UB exhibits great accuracy for a wide range of SE values and, thus, can be utilized for explicit assessment of the influence of SE on EE and for analytically addressing the EE optimization problems. Using this tight EE UB, our analysis unfolds two EE optimization issues: Given the number of transmit and receive antennas, an optimum value of SE is derived, such that the overall EE can be maximized, and given a specific value of SE, the optimal number of antennas is derived for maximizing the system EE.
Article
Full-text available
Assume that a multi-user multiple-input multiple-output (MIMO) communication system must be designed to cover a given area with maximal energy efficiency (bits/Joule). What are the optimal values for the number of antennas, active users, and transmit power? By using a new model that describes how these three parameters affect the total energy efficiency of the system, this work provides closed-form expressions for their optimal values and interactions. In sharp contrast to common belief, the transmit power is found to increase (not decrease) with the number of antennas. This implies that energy efficient systems can operate at high signal-to-noise ratio (SNR) regimes in which the use of interference-suppressing precoding schemes is essential. Numerical results show that the maximal energy efficiency is achieved by a massive MIMO setup wherein hundreds of antennas are deployed to serve relatively many users using interference-suppressing regularized zero-forcing precoding.
Article
Full-text available
Along with spectral efficiency (SE), energy efficiency (EE) is becoming one of the key performance evaluation criteria for communication system. These two criteria, which are conflicting, can be linked through their trade-off. The EE-SE tradeoff for the multi-input multi-output (MIMO) Rayleigh fading channel has been accurately approximated in the past but only in the low-SE regime. In this paper, we propose a novel and more generic closed-form approximation of this trade-off which exhibits a greater accuracy for a wider range of SE values and antenna configurations. Our expression has been here utilized for assessing analytically the EE gain of MIMO over single-input single-output (SISO) system for two different types of power consumption models (PCMs): the theoretical PCM, where only the transmit power is considered as consumed power; and a more realistic PCMaccounting for the fixed consumed power and amplifier inefficiency. Our analysis unfolds the large mismatch between theoretical and practical MIMO vs. SISO EE gains; the EE gain increases both with the SE and the number of antennas in theory, which indicates that MIMO is a promising EE enabler; whereas it remains small and decreases with the number of transmit antennas when a realistic PCM is considered.
Article
Full-text available
In order to quantify the energy efficiency of a wireless network, the power consumption of the entire system needs to be captured. In this article, the necessary extensions with respect to existing performance evaluation frameworks are discussed. The most important addenda of the proposed energy efficiency evaluation framework (E3F) are a sophisticated power model for various base station types, as well as large-scale long-term traffic models. The BS power model maps the RF output power radiated at the antenna elements to the total supply power of a BS site. The proposed traffic model emulates the spatial distribution of the traffic demands over large geographical regions, including urban and rural areas, as well as temporal variations between peak and off-peak hours. Finally, the E3F is applied to quantify the energy efficiency of the downlink of a 3GPP LTE radio access network. The paper received the 2014 IEEE Communications Society Fred W. Ellersick Prize, awarded to honor outstanding articles published in a Communications Society magazine in the previous three calendar years.
Article
Full-text available
The dramatic increase of network infrastructure comes at the cost of rapidly increasing energy consumption, which makes optimization of energy efficiency (EE) an important topic. Since EE is often modeled as the ratio of rate to power, we present a mathematical framework called fractional programming that provides insight into this class of optimization problems, as well as algorithms for computing the solution. The main idea is that the objective function is transformed to a weighted sum of rate and power. A generic problem formulation for systems dissipating transmit-independent circuit power in addition to transmit-dependent power is presented. We show that a broad class of EE maximization problems can be solved efficiently, provided the rate is a concave function of the transmit power. We elaborate examples of various system models including time-varying parallel channels. Rate functions with an arbitrary discrete modulation scheme are also treated. The examples considered lead to water-filling solutions, but these are different from the dual problems of power minimization under rate constraints and rate maximization under power constraints, respectively, because the constraints need not be active. We also demonstrate that if the solution to a rate maximization problem is known, it can be utilized to reduce the EE problem into a one-dimensional convex problem.
Article
Full-text available
Wireless communication systems based on multiple-input multiple-output (MIMO) technology have the potential to achieve increased spectral efficiency with no additional transmit power or bandwidth requirements. This can be achieved by exploiting the spatial subchannels between the multiple transmit and receive elements. The enhanced performance of MIMO technology makes it a strong candidate for ad hoc systems requiring very high throughput, such as high-definition multimedia streaming in the home. In this article we compare the performance of conventional and newly proposed MIMO architectures for such applications. Specifically designed antenna arrays are proposed as a solution to significantly reduced MIMO performance in line-of-sight (LoS) conditions. To help in the design of such systems, the criterion for achieving maximum MIMO capacity in LoS is presented as a function of the transmitter-to-receiver separation distance, the array orientation and the element spacings. The performance of the LoS-enhanced proposal and the conventional system is quantified in terms of the channel capacity, the packet error rate (PER) and the throughput by employing a MIMO-enabled WLAN physical layer simulator. The results from MIMO measurements performed inside an anechoic chamber and in an indoor environment are presented to validate the theoretic predictions
Article
We investigate the use of multiple transmitting and/or receiving antennas for single user communications over the additive Gaussian channel with and without fading. We derive formulas for the capacities and error exponents of such channels, and describe computational procedures to evaluate such formulas. We show that the potential gains of such multi-antenna systems over single-antenna systems is rather large under independenceassumptions for the fades and noises at different receiving antennas.
Article
In this paper, we investigate spatial multiplexing at millimeter (mm) wave carrier frequencies for short-range indoor applications by quantifying fundamental limits in line-of-sight (LOS) environments and then investigating performance in the presence of multipath and LOS blockage. Our contributions are summarized as follows. For linear arrays with constrained form factor, an asymptotic analysis based on the properties of prolate spheroidal wave functions shows that a sparse array producing a spatially uncorrelated channel matrix effectively provides the maximum number of spatial degrees of freedom in a LOS environment, although substantial beamforming gains can be obtained by using denser arrays. This motivates our proposed mm-wave MIMO architecture, which utilizes arrays of subarrays to provide both directivity and spatial multiplexing gains. System performance is evaluated in a simulated indoor environment using a ray-tracing model that incorporates multipath effects and potential LOS blockage. Eigenmode transmission with waterfilling power allocation serves as a performance benchmark, and is compared to the simpler scheme of beamsteering transmission with MMSE reception and a fixed signal constellation. Our numerical results provide insight into the spatial variations of attainable capacity within a room, and the combinations of beamsteering and spatial multiplexing used in different scenarios.
Article
This paper investigates the ergodic capacity of Ricean-fading multiple-input-multiple-output (MIMO) channels with rank-1 mean matrices under the assumption that the channel is unknown at the transmitter and perfectly known at the receiver. After introducing the system model and the concept of ergodic capacity of MIMO channels, we derive the explicit expressions for the expected values of the determinant and log-determinant of complex noncentral Wishart matrices. Subsequently, we obtain new upper and lower bounds on the ergodic capacity of rank-1 Ricean-fading MIMO channels at any signal-to-noise ratio (SNR). We show that our bounds are tighter than previously reported analytical bounds, and discuss the impact of spatial fading correlation and Ricean K-factor with the help of these bounds. Furthermore, we extend the analysis of ergodic capacity to frequency selective spatially correlated Ricean-fading MIMO channels. We demonstrate that the calculation of ergodic capacity of frequency selective fading MIMO channels can be converted to the calculation of the one of equivalent frequency flat-fading MIMO channels. Finally, we present numerical results that confirm the theoretical analysis
Article
We investigate the use of multiple transmitting and/or receiving antennas for single user communications over the additive Gaussian channel with and without fading. We derive formulas for the capacities and error exponents of such channels, and describe computational procedures to evaluate such formulas. We show that the potential gains of such multi-antenna systems over single-antenna systems is rather large under independence assumptions for the fades and noises at different receiving antennas. 1 Introduction We will consider a single user Gaussian channel with multiple transmitting and/or receiving antennas. We will denote the number of transmitting antennas by t and the number of receiving antennas by r. We will exclusively deal with a linear model in which the received vector y 2 C r depends on the transmitted vector x 2 C t via y = Hx+ n (1) where H is a r Theta t complex matrix and n is zero-mean complex Gaussian noise with independent, equal variance real and imaginary p...