412 IEEE WIRELESS COMMUNICATIONS LETTERS, VOL. 1, NO.

5, OCTOBER 2012

Opportunistic Energy Efficient Cooperative Communication

Osama Amin, Member, IEEE, and Lutz Lampe, Senior Member, IEEE

Abstract—Inspired by the green communication trend of

next-generation wireless networks, we present an opportunistic
energy-efficient cooperative communication method. Different
from classical relaying selection schemes, which adopt either end-
to-end signal-to-noise ratio or capacity as selection metrics, our
opportunistic relaying scheme uses an energy efficiency metric for
selecting the best relay or resorting to direct transmission. Simu-
lation results demonstrate the significant energy savings achieved
by our scheme when compared with opportunistic decode-and-
forward based on capacity. In addition to that, dynamic time
allocation and increasing the number of participating relays are
shown to have a notable effect on energy efficiency.
Index Terms—Cooperative communications, energy efficiency
(EE), opportunistic decode-and-forward (ODF), green communi-
Fig. 1. Multiple relay cooperative network.
cation, multiple relays, relay selection.

I. I NTRODUCTION dynamic time allocation between the decode-and-forward (DF)

transmission phases to maximize EE while respecting rate
T HE tremendous proliferation of wireless devices and ser-
vices has significantly increased the energy consumption
of wireless communications technology and its contribution to
requirements. We demonstrate the superiority of the proposed
system in improving EE through relevant simulation examples.
global CO2 emissions. For example, mobile communication
uses huge amounts of diesel to generate the required electrical II. S YSTEM M ODEL
energy for operating base stations [1]. Towards cutting down We assume a multiple-relay cooperative network which
the carbon footprint effect, different international projects have consists of a single source (S), a single destination (D) and L
been launched with the goal of improving energy efficiency relays Ri as shown in Figure 1. All nodes are equipped with a
(EE) of wireless communication networks[2]. single antenna, and the relays are assumed to work with half-
Cooperative communication is a powerful concept for ex- duplex DF mode. The transmission-link channels are modelled
tending coverage and improving EE without including extra as quasi static Rayleigh fading channels, where the channels
infrastructure [3]. Opportunistic decode-and-forward (ODF) are constant during the transmission of a block of symbols,
and relay selection are two cooperative communication tech- with amplitude hXY , X,Y ∈ {S, Ri , D}, i = 1, 2, ..., L. The
niques that have been widely investigated in the literature, path loss of the X − Y link is modeled as
e.g., [4], [5]. ODF switches between direct transmission (DT) GXY = (4πd0 /λ)−2 (dXY /d0 )−κ , (1)
and relaying depending on a performance metric such as
achievable rate [4], and relay selection chooses the best among where dXY is the distance between X and Y, d0 is a reference
a set of available relays based on, e.g., signal-to-noise ratio distance, λ is the wavelength and κ is the path loss exponent
(SNR) of the cooperative links [5]. [6].
In this letter, we follow the recent trend towards (coop- The cooperative scheme elects to operate either with the
erative) communication systems that support the green vi- help of one of the available relays in the network using the
sion of next-generation wireless networks [2] and present an DF cooperative mode or without cooperation using only the
opportunistic energy-efficient (OEE) relay selection coopera- DT mode. In case of the DF cooperative mode, we apply
tive communication scheme. OEE relaying extends the ODF the transmission schedule referred to as Protocol II in [7].
scheme to the multiple-relay scenario based on an EE metric. This DF protocol divides the transmission time between two
The EE metric considers the energy consumption in all circuits phases, namely the broadcasting and relaying phases. In the
of the participating nodes and thus can avoid nodes with broadcasting phase the source broadcasts the data to the
inefficient energy usage. Furthermore, our system considers network, and in the relaying phase, a selected relay forwards
the data to the destination. In the DT mode, the source
Manuscript received March 20, 2012. The associate editor coordinating the uses the whole transmission time for sending the data to the
review of this letter and approving it for publication was A. Bletsas.
The authors are with the Department of Electrical and Computer Engineer- destination.
ing, University of British Columbia, Vancouver, BC, Canada (e-mail: {oamin, The selection principle of the proposed system (i.e., OEE)
lampe}@ece.ubc.ca). is based on an EE criterion which includes the effect of two
This work was supported by the National Sciences and Engineering
Research Council (NSERC) of Canada. classical metrics: transmitted energy consumption and end-
Digital Object Identifier 10.1109/WCL.2012.061212.120206 to-end capacity. In addition to that, it considers the impact
2162-2337/12$31.00 
c 2012 IEEE

Authorized licensed use limited to: AMITY University. Downloaded on July 05,2020 at 20:00:16 UTC from IEEE Xplore. Restrictions apply.
AMIN and LAMPE: OPPORTUNISTIC ENERGY EFFICIENT COOPERATIVE COMMUNICATION 413

of circuit-energy consumption, which has not been considered where the last constraint for 0 < ρ < 1 ensures that
previously for relay selection methods [8], [9], [10], [2]. Under improvements in energy efficiency do not incur an arbitrary
the assumption of dynamic time allocation, a fraction t and degradation of the end-to-end rate. After solving (7) for
(1 − t) of the transmission time is spent for the broadcasting m = 1, . . . , L, and considering EE0 from (4), the relay with
and relaying phase, respectively, to maximize EE and meet the maximal EE metric or DT is chosen for transmission. In
minimal rate constraints, see Figure 1. the following, we consider the optimization problem (7) with
and without the capacity constraint (7c) to investigate possible
III. OEE R ELAY S ELECTION S CHEME rate losses due to energy-efficient relay selection.
First, we observe that using (6), the EE expression for the
Using the channel states of the network links, the OEE
mth relay defined in (5) can be written as a piecewise of
relaying scheme selects between DT or DF modes based on
EEm,1 =
argmax (EEm ) , (2) tm CSRm
m=0,1,··· ,L ,
(1 + α) (tm PS + (1 − tm ) PRm ) + (tm + 1) Pcr + Pct
where m = 0 denotes the DT mode and m = i, i = 0, denotes
0 < tm ≤ tm,joint , (8)
the DF mode that uses the ith relay for cooperation. The EE
of the DT mode is expressed similar to [8] as and
CSD EEm,2 =
EE0 = , (3)
(1 + α) PS + Pct + Pcr tm CSD + (1 − tm )CRm D
,
where CSD is the channel capacity of the S − D link, α is (1 + α) (tm PS + (1 − tm ) PRm ) + (tm + 1) Pcr + Pct
a constant that depends on the drain efficiency of the power tm,joint ≤ tm < 1 . (9)
amplifier and the peak-to-average power ratio [9], Pct is the
transmitter-circuit power consumption, Pcr is the receiver- EEm,1 is monotonically increasing in tm , and EEm,2 is
circuit power consumption, and PS is the transmitted power monotonically decreasing if
 
of the source. CSD is expressed as (1 + α) (PS + PRm ) + Pcr
  CSD < CRm D 1 + , (10)
GSD h2SD PS (1 + α) PRm + Pcr + Pct
CSD = B log2 1 + , (4)
N0 Nf B and monotonically increasing otherwise. The breakpoint
tm,joint follows from (6) as
where N0 is the noise power spectral density, Nf is the receiver
CRm D
noise figure and B is the bandwidth. Under the assumption of tm,joint = . (11)
identical circuit-power consumption for the source, the relays CSRm + CRm D − CSD
and the destination, the EE of the DF cooperative mode using tm,joint exists in the interval described by constraint (7b), i.e.
the mth relay, m ≥ 1, can be written as tm,joint ∈ (0, 1), if and only if
EEm = C1 : CSD < CSRm . (12)
Cm Both the location of tm,joint and the behavior of EEm,2 ,
,
(1 + α)(tm PS + (1 − tm )PRm ) + (tm + 1)Pcr + Pct whether it is monotonically increasing or decreasing, play
m = 1, 2, . . . , L, (5) a significant role in determining the relaying mode in OEE
relaying. In the following, we consider the possible scenarios
where tm is the fraction of time for the broadcasting phase for problem (7), first without capacity constraint (7c).
when the mth relay is used for cooperation, PRm is the • Scenario I: When C1 is not satisfied, DT is preferable
transmit power of this relay, and Cm is expressed as [4] over using relay m. This is illustrated in Figure 2(a),
Cm = min (tm CSRm , tm CSD + (1 − tm )CRm D ) , together with the associated EE and capacity metrics.
• Scenario II: If C1 is satisfied, i.e., tm,joint ∈ (0, 1)
m = 1, 2, . . . , L. (6)
and EEm,2 is monotonically increasing in tm , then DT
We note that tm is only defined for m ≥ 1, when the mth achieves a higher EE than DF with relay m. This is
relay is used for DF cooperation. The expressions for the link illustrated in Figure 2(b).
capacities CSRm and CRm D are the same as in (4), using the • Scenario III: If C1 is satisfied, i.e., tm,joint ∈ (0, 1)
corresponding channel gains and transmit powers. In previous and EEm,2 is monotonically decreasing in tm , then DF
works, the time allocation ratio tm has been chosen either fixed with relay m achieves a higher EE than DT if
as 0.5 or adjusted dynamically between 0 and 1 to improve Pcr CSD
transmitted power consumption or channel capacity [4]. In C2 : CSRm > CSD +
(1 + α) PS + Pct + Pcr
our work, tm is computed dynamically to improve the EE  
(1 + α) PRm + Pcr + Pct CSRm − CSD
metric, which takes power consumption and system capacity + CSD .
(1 + α) PS + Pct + Pcr CRm D
into account. Hence, given a relay m, tm is obtained from
(13)
max EEm (7a)
tm In this scenario, maximum EE is achieved at time allo-
subject to 0 <tm < 1 (7b) cation1 tm,UC = tm,joint with end-to-end rate Cm,UC .
Cm ≥ ρCSD , (7c) 1 “UC” stands for “unconstrained” as constraint (7c) is not considered here.

Authorized licensed use limited to: AMITY University. Downloaded on July 05,2020 at 20:00:16 UTC from IEEE Xplore. Restrictions apply.
414 IEEE WIRELESS COMMUNICATIONS LETTERS, VOL. 1, NO. 5, OCTOBER 2012

4 Scenario I 4 Scenario II 4 Scenario IV (C >C ) 4 Scenario IV (C <C )

x 10 x 10 x 10 SD R D x 10 SD R D
m m
10 6 4 3
EEm EEm EEm EE
m
9 EE0 EE0 EE0 EE0
C 5 C Cm Cm
m m
8
CSD CSD CSD
3 CSD
7
4 2
6

5 3 2

4
2 1
3
1
2
1
1

0 0 0 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
t t t t
tm, joint tm, joint tm, joint
(a) (b) (a) (b)

Fig. 2. The relations of EE and channel capacities of the DT and cooperative Fig. 4. The relations of EE and channel capacities of the DT and cooperative
DF modes versus time allocation for Scenarios I (left) and II (right). System DF modes versus time allocation for Scenario IV. System parameters from
parameters from Table I, and a single channel realization is considered. Table I, and a single channel realization is considered.

4 Scenario III (C >C ) 4 Scenario III (CSD < CR D)

x 10 SD R D x 10 m
m
5
EEm EEm satisfies C1 and C2, DT is performed.
EE
0
EE
0 We now turn to (7) when the capacity constraint (7c) is
3
Cm
4
Cm included. Then, only Scenario III with CSD > CRm D is
CSD CSD
affected, since only in this case the DF mode achieves a better
EE than the DT mode at the cost of a rate loss (see Figure 3).
3
2 As can be seen from Figure 3(a), the optimal time allocation is
identical to tm,UC if the corresponding end-to-end rate Cm,UC
2 C
meets the capacity constraint, otherwise the time allocation is
m,UC
=Cm,C increased to adjust the end-to-end rate. This rate compensation
Cm,UC 1 for the DF mode incurs an EE loss as shown in Figure 3(a).
1 In particular, EE of DF is larger than EE of DT only if
tm ,C
 
CSD − CRm D
Cm,C= ρCSD
CRm D
C3 : ρ < + ×
0 0 CSD CSD
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
tm,UC t tm,UC=tm,C t (1 + α) (PS CRm D − PRm CSD ) + (Pcr + Pct ) (CRm D − CSD )
(a) (b) .
((1 + α)PS + Pcr + Pct ) CRm D − (Pct + (1 + α)PRm ) CSD
Fig. 3. The relations of EE and channel capacities of the DT and cooperative (14)
DF modes versus time allocation for Scenario III. The shaded area in Figure
3(a) illustrates the adjustment of time allocation tm until capacity constraint
The optimal time allocation ratio for the capacity constrained
(7c) is met. System parameters from Table I, and a single channel realization problem is expressed as
is considered. 
tm,UC Cm,UC ≥ ρCSD
tm,C = ρCSD −CRD . (15)
CSD −CRD Cm,UC < ρCSD
Figure 3 (disregard tm,C and Cm,C for now) illustrates Hence, in the capacity-constrained problem (7), all relays for
this case, with Figure 3(a) for CSD > CRm D and Figure which C1, C2, and C3 are satisfied are determined and the EE
3(b) for CSD < CRm D . We note that in the latter case for these relays is computed at tm,C defined in (15).
both EE and capacity based ODF choose relaying over
DT. The selected relay in the case of multiple relays may IV. S IMULATION R ESULTS
be different for EE and capacity based ODF though. In this section, we compare the performance of the OEE
• Scenario IV: The final scenario is when tm,joint ∈ scheme with that of ODF based on capacity, and we evaluate
(0, 1), EEm,2 is monotonically decreasing and C2 is the effect of number of relays and dynamic time allocation on
not satisfied. Then DT achieves a better EE than DF. EE. For this purpose, we use the system parameters for the
This is shown in Figure 4. We observe from Figure 4(b) 2.5 GHz radio from [9] summarized in Table I.
that capacity based ODF may choose DF over DT when In the first set of results, we study the tradeoff between EE
OEE does not. and end-to-end rate when using OEE relaying and compare
Since when C2 is true, also (10) is satisfied, OEE relaying it with that for capacity-based ODF. For this purpose, a
chooses DF over DT if and only if C1 and C2 are met. Hence, single relay located in the middle distance between source
OEE relaying determines all relays for which C1 and C2 are and destination is considered, and Figure 5 shows the OEE-
true, and then selects the relay with the largest EE. If no relay vs-ODF scatter plot obtained from 106 channel realizations.

Authorized licensed use limited to: AMITY University. Downloaded on July 05,2020 at 20:00:16 UTC from IEEE Xplore. Restrictions apply.
AMIN and LAMPE: OPPORTUNISTIC ENERGY EFFICIENT COOPERATIVE COMMUNICATION 415

TABLE I
S IMULATION PARAMETERS . 30
L=1
Pcr = 112 mW Pct = 98 mW α = 1.88 L=2
Nf = 10 dB N0 = −171 dBm/Hz B = 10 kHz 25
L=3
κ = 3.5 d0 = 1 λ = 0.12 L=4
PS = 20 dBm PRm = 5 dBm dSD = 500 m

Energy savings in percent

20 L=5

15

10

0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
r

Fig. 6. Energy savings when using the OEE scheme with dynamic time
allocation compared to OEE with fixed time allocation tm = 0.5. The L
relays are located on the direct line between S and D with distance rdSD
from the source.

which can mostly be redeemed, however, including a capacity

constraint in the dynamic time allocation problem.
Fig. 5. Scatter plot of the EE and the end-to-end rate for capacity-based
ODF and the OEE scheme assuming unconstrained and constrained capacity
problems (7). R EFERENCES
[1] C. Han et al., “Green radio: radio techniques to enable energy-efficient
wireless networks,” IEEE Commun. Mag., vol. 49, no. 6, pp. 46–54,
For the unconstrained capacity problem (top row in Figure June 2011.
[2] Y. Chen, S. Zhang, S. Xu, and G. Li, “Fundamental trade-offs on green
3), the scatter plot shows that the EE improvement of the wireless networks,” IEEE Commun. Mag., vol. 49, no. 6, pp. 30–37,
OEE scheme comes with a corresponding degradation in the June 2011.
end-to-end rate. This observation emphasizes that EE metrics [3] F. H. P. Fitzek and M. D. Katz, Cooperation in Wireless Networks:
Principles and Applications; Real Egoistic Behavior is to Cooperate!
need to be applied carefully when end-to-end rate is important Springer–Verlag, 2006.
too. This is accomplished through the capacity-constrained [4] D. Gunduz and E. Erkip, “Opportunistic cooperation by dynamic
problem, whose results are shown in the bottom row of Figure resource allocation,” IEEE Trans. Wireless Commun., vol. 6, no. 4, pp.
1446–1454, Apr. 2007.
5 for ρ = 0.9. We observe that the gains in EE of OEE [5] A. Bletsas, A. Khisti, D. Reed, and A. Lippman, “A simple cooperative
over conventional ODF are only little affected, while the rate diversity method based on network path selection,” IEEE J. Sel. Areas
losses are much reduced. Hence, the benefits of improved EE Commun., vol. 24, no. 3, pp. 659–672, Mar. 2006.
[6] V. Erceg, L. Greenstein, S. Tjandra, S. Parkoff, A. Gupta, B. Kulic,
outweigh the occasional reduction in rate (in addition to the A. Julius, and R. Bianchi, “An empirically based path loss model
fact that EE was deemed to be the decisive figure of merit). for wireless channels in suburban environments,” IEEE J. Sel. Areas
The second set of results illustrates the role of dynamic Commun., vol. 17, no. 7, pp. 1205–1211, July 1999.
[7] R. U. Nabar, H. Bölcskei, and F. W. Kneubühler, “Fading relay channels:
time allocation for the proposed OEE scheme. We consider performance limits and space-time signal design,” IEEE J. Sel. Areas
a network with L = 1, . . . , 5 relays. Figure 6 shows the Commun., vol. 22, no. 6, pp. 1099–1109, Aug. 2004.
energy for different numbers of relays versus the location of [8] G. Miao, N. Himayat, Y. Li, and D. Bormann, “Energy efficient design
in wireless OFDMA,” in Proc. 2008 IEEE International Conference on
the relays. For simplicity, all relays are assumed to be located Communications, pp. 3307–3312.
on the direct line between S and D with distance rdSD from the [9] S. Cui, A. Goldsmith, and A. Bahai, “Energy-constrained modulation
source. As can be seen from Figure 6, dynamic time allocation optimization,” IEEE Trans. Wireless Commun., vol. 4, no. 5, pp. 2349–
2360, Sep. 2005.
plays a particularly critical role for EE especially when the [10] T. Chen, H. Kim, and Y. Yang, “Energy efficiency metrics for green
relays are located close to the middle of the S − D distance, wireless communications,” in Proc. 2010 International Conference on
and significant gains in EE are possible through the suggested Wireless Communications and Signal Processing, pp. 1–6.
adaptation.

V. C ONCLUSION
In this letter, we proposed the OEE relay selection coop-
erative communication scheme. Towards the goal of green
communication systems, the OEE scheme considers EE as the
primary figure of merit. The improvement in EE over classical
ODF relaying comes at the price of reduced end-to-end rate

Authorized licensed use limited to: AMITY University. Downloaded on July 05,2020 at 20:00:16 UTC from IEEE Xplore. Restrictions apply.