Professional Documents
Culture Documents
Dynamic Subchannel and Power Allocation in OFDMA-Based DF Cooperative Relay Networks
Dynamic Subchannel and Power Allocation in OFDMA-Based DF Cooperative Relay Networks
Dynamic Subchannel and Power Allocation in OFDMA-Based DF Cooperative Relay Networks
Abstract— This paper focuses on subchannel and power al- two kinds of optimization techniques which are rate adaptive
location problem in OFDMA-based decode-and-forward (DF) (RA) and margin adaptive (MA). In RA problem, the objective
cooperative relay networks. The objective of this problem is is to maximize users’ error free capacity with the constraints
to maximize the total data rate under the constraints of joint
total transmission power and subchannels occupation, while of total transmission power and co-channel interference (CCI)
maintaining the maximum fairness among multiple relay nodes. avoidance [8] and [9]. While the objective of MA is to achieve
We transform this mixed integer programming problem to a the minimum overall transmit power given the constraints on
standard convex optimization one through Lagrangian relaxation the users’ data rates or bit error rates (BER). An adaptive
method. Accordingly, the optimal solution can be achieved by subcarrier, bit, and power allocation algorithm was proposed
using Karush-Kuhn-Tucker (KKT) conditions. In addition, we
propose two suboptimal subchannel allocation schemes with and in [10], whose objective is to minimize the total transmit
without subchannel permutation (SP), which reduce significantly power. In [11], Kim et al. solved the optimal subcarrier and
the computational complexity with small performance loss. The bit allocation problem by using integer programming(IP), and
simulation results show that the performance of the proposed proposed linear-programming(LP)-based suboptimal algorithm
suboptimal algorithms with SP approaches asymptotically that which is considerably simpler to implement than the optimal
of the optimal algorithm and with the increment of the number
of relay nodes. IP.
All the aforementioned work, which investigated different
I. I NTRODUCTION power, bit and subchannel allocation schemes with diverse
Resource allocation such as power control and frequency optimization objectives to improve the system performance,
allocation has been regarded as one of effective ways to im- focused on peer-to-peer wireless OFDM systems. However,
prove the capacity of wireless networks. In resource allocation, there are still many unresolved problems of resource allocation
dynamic subcarrier and power allocation (SPA) can constantly in cooperative relay networks. It is all known that OFDM can
maintain high transmission rate. Thus, intense interest has been eliminate inter-symbol interference by dividing the available
focused on the resource allocation, especially in cooperative bandwidth into multiple orthogonal subchannels. Hence, Lin
OFDMA networks. Dai et al. [12] applied the technique to cooperative system
The main idea of cooperative communication is to achieve and proposed an OFDMA-based selective relaying scheme,
spatial diversity through distributed cooperating terminals. In where the relay selection at each hop is performed on a per-
the pioneer work on cooperative transmission mechanism [1], subcarrier basis. By assigning the subcarriers to different relay
a CDMA-based two users cooperative transmission framework nodes, the error probability of the whole OFDM symbol is
was adopted to demonstrate the mechanism and advantages of greatly reduced. Based on this idea, the power allocation over
cooperative diversity. Several efficient cooperative protocols frequency subchannels at source and relay nodes in OFDM
and their outage behaviors were analyzed in [2]. Generally, co- cooperative networks was studied in [13] and [14]. However,
operation protocol can be cataloged into two classes, i.e., fixed they only considered the scenario with the AF cooperation
relay protocol and selective relay protocol [2]. The former protocol and a single relay node. In OFDMA-based DF
one includes amplify-and-forward (AF), decode-and-forward cooperative relay networks, Bo Gui et al. [15] proposed
(DF) [3], compress-and-forward (CF) [4], [5]. The later one several bit loading and power allocation schemes by using
comprises selection decode-and-forward, coded cooperation an equivalent channel gain, which aimed at minimizing the
[6], and dynamic decode-and-forward [7]. Most research on total transmission power under a target rate constraint. In [16],
cooperation protocol exploit a two-stage transmission scheme. the power management problem over a multiuser OFDMA-
In the first stage, the source node transmits and the relay and based cooperative network with the objective of minimizing
destination nodes listen, while in the second stage, the relay the total transmission power under the constraint of each user’s
nodes transmit the received message to the destination node rate was analyzed. By dividing the total transmission power
and the source node is idle. The optimization of resource into two parts according to the sizes of sub-frames, separate
allocation in OFDM systems offers substantial gains to the transmission power minimization can be done in [17], which
system performances [8]- [11]. In resource allocation, there are is subject to a requested data rate on each link.
⎧
⎪
⎪ Psrk (n)Gsrk (n)
1 ⎨ log2 1 + ,
N0
ĂĂ
k=1
N
N
where ρk,n is an indicator which denotes whether or not ρk (n, j) Psrk (n)Gsrk (n)
subcarrier n is used at relay node k. Constraint C1 indicates s.t. C1 z log2 1 + ,
n=1 j=1
2 N0 ρk (n, j)
that the total transmission power for an OFDM symbol con- ⎛ ⎞
sists of two parts, and C2 indicates each subcarrier n can only
K N N
G (n)
Psrk (n) ⎝1 + ⎠ PT ,
srk
be used by at most one relay at each stage. C2. ρk (n, j)
G (j)
As we can see from (3), the capacity of subchannel n n=1 j=1 rk d
k=1
30
The total minimum transmission power that should be allo-
cated to subchannel n is 25
20
PT (n) = Psrk (n) + Prk d (n),
15
Preq (n)
= . (12)
Gsrk d (n) 10
5
where Gsrk d (n) is defined as the equivalent channel gain of
the subchannel occupied by the relay node Rk , which is given 0
0 2 4 6 8 10 12 14 16 18 20
Transmit Power ( R=8 )
as
Fig. 2. Average total capacity for different SNR
Gsrk (n)Grk d (n)
Gsrk d (n) = . (13)
Gsrk (n) + Grk d (n)
c) update Rk according to (9) by using the result of
As we can see, (13) is equivalent to the harmonic mean of
b), S = S − n and S = S − j.
the gains of subchannel n occupied by the kth relay in the
step 3 : While S = φ, S = φ
two transmission stages.
We exploit the subcarrier with the largest Gsrk d (n) at a) find k satisfying Rk ≤ Rk for all k , 0 ≤ k ≤ K,
relay k. Hereby, the subchannel allocation procedure can be
described as b) for the found k, pick out the n satisfying
|hk,n | ≥ |hk,n | and j satisfying |hk,j | ≥ |hk,j |
step 1 : Initialization
for all n ∈ S and j ∈ S ,
set Rk = 0 for k = 1, 2, ..., K, S = {1, 2, ..., N }.
step 2 : For k = 1 to K
c) update Rk , S and S with the k, n and j, respec-
a) calculate the equivalent channel gain according to tively:
(13) for all n ∈ S, and find Rk = Rk + C(hk,n ), S = S − {n}, S = S − {j}.
n∗ = arg max Gsrk d (n),
n∈S V. S IMULATION R ESULTS
b) for the found n∗ , compare |hsrk (n∗ )| with
|hrk d (n∗ )|, choose the smaller one and update Rk In this section, we present some simulation results to
according to (9): compare the performance of different subcarrier and power
Rk = C(hk,n ), S = S − {n}. allocation algorithms. We employ an OFDMA-based cooper-
ative communication network with a single source-destination
step 3 : While S = φ (empty set)
pair and K relay nodes. The relay nodes are independently
a) find k satisfying Rk ≤ Rj for all j, 0 ≤ j ≤ K, and identically distributed in the area between the source and
relay nodes. We also assume that the total subcarrier number
b) for the found k, repeat step 2, N is fixed to be 64 and the power spectral density of AWGN
c) update Rk , S and S : is normalized to be 1.
Rk = Rk + C(hk,n ), S = S − {n}. In Fig.2, we compare the average total data rate of the
B. Subchannel Allocation with SP optimal algorithm, suboptimal algorithm without and with SP.
In this experiment, the number of relay nodes is fixed to be
The subcarrier allocation algorithm with SP can be de- 8. From Fig.2 we can see that with the increment of transmit
scribed as power, the performance of the proposed suboptimal approach
step 1 : Initialization asymptotically to the optimal one with SP. This is because the
set Rk = 0 for k = 1, 2, ..., K, S = {1, 2, ..., N } and subchannel allocation algorithm utilizes the subchannels with
S = {1, 2, ..., N }. large SNR as much as possible. We also observe that both
step 2 : For k = 1 to K of this two algorithms achieve more performance gain than
a) find n satisfying |hsrk (n)| ≥ |hrk d (n )| and j the suboptimal algorithm without SP. It can be found that the
satisfying |hsrk (j)| ≥ |hrk d (j ) for all n, n ∈ S averaged total data rate increases more and more quickly at
and j, j ∈ S , high transmit power. This result validates the conclusion in [1]
b) compare |hsrk (n)| with |hrk d (j)|, and select the that the cooperation scheme can greatly improve the achievable
smaller one, rate region when the channels between the source and relay
nodes are better.
3.5 implementation.
3 ACKNOWLEDGEMENT
2.5
This work is supported by the National Natural Science
optimal Foundation of China (No. 60572157). We thanks Feng She
2 equal power with SP for discussion about the convex optimization problem.
equal power without SP
1.5
R EFERENCES
1 [1] A. Sendonaris, E. Erkip and B. Aazhang, “User Cooperation Diversity-
part I and II: System Description,” IEEE Trans. Commun., vol.51, no. 11,
0.5
2 4 6 8 10 12 14 16
pp.1927-1948, Nov. 2003.
The number of Relay Node [2] J. N. Laneman, D. N. C. 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.
Fig. 3. Average total capacity for different relay number K [3] A. Nosratinia, T. Hunter, and A. Hedayat, “Cooperative communication
in wireless networks,” IEEE Commun. Mag., vol. 42, no. 10, pp. 68-73,
Oct. 2004.
[4] T. M. Cover, A.A. El Gamal, “Capacity theorems for the relay channel”,
Fig.3 illustrates the data rate with different number of relay IEEE Trans. Inf. Theory, vol. IT-25, No 5, pp 572-584, Sep. 1979.
nodes. In this experiment, we compare the date rate of the [5] G. Kramer., M. Gastpar and P. Gupta, “Cooperative Strategies and
proposed algorithms. The comparison is done at the SNR Capacity Theorems for Relay Networks,” IEEE Trans. Inf. Theory, vol.
51, no. 9, pp. 3037-3063, Sep. 2005.
level 5dB. From Fig.3, we can observe that with increment [6] T. E. Hunter and A. Nosratinia, “Cooperative diversity through coding,”
of the number of relay nodes, the data rate can be improved. in Proc. IEEE International Symposium on Information Theory (ISIT),
However, the slope of the curves become small when the user Laussane, Switzerland, p. 220, Jul. 2002.
[7] K. Azarian, H. El Gamal and P. Schniter, “On the Achievable Diversity-
number is large, which indicates that the incremental gains Multiplexing Tradeoff in Half-Duplex Cooperative Channels,” IEEE
from additional relay nodes will diminished as the number of Trans. Info. Theory, vol. 51, no. 12, pp. 4152-4172, Dec. 2005.
the relay nodes grows. This conclusion is consistent with that [8] W. Rhee and J. M. Cioffi, “Increasing in capacity of multiuser OFDM
system using dynamic subchannel allocation,” in Proc. IEEE Int. Vehic-
in [1]. We also find that the performance of the proposed sub- ular Tech. Conf., vol. 2, pp. 1085¨C1089, Tokyo, Japan, May 2000.
optimal algorithm with SP approaches asymptotically that of [9] Z. K. Shen, J. G. Andrews and B. L. Evans, “Adaptive resource allocation
the optimal algorithm. This is because the power distribution in multiuser OFDM systems with proportional rate constraints,” IEEE
Trans. Wireless Commun., vol. 4, no. 6, pp. 2726-2736, Nov. 2005.
among the relay nodes tends to be uniform when the number [10] C. Y. Wong, R. S. Cheng, K. B. Letaief, and R. D. Murch, “Multicarrier
of relay nodes is large, which is similar to the assumption OFDM with adaptive subcarrier, bit, and power allocation,” IEEE J. Sel.
of the suboptimal algorithm with SP. We also find that the Areas Commun., vol. 17, no. 10, pp. 1747¨C1758, Oct. 1999.
[11] I. Kim I. Park and Y. H. Lee, “Use of Linear Programming for
performance of the suboptimal algorithm without SP is away Dynamic Subcarrier and Bit Allocation in Multiuser OFDM,” IEEE Trans.
from that of the algorithms with SP when K is large. This Vehicular Tech., vol. 55, no. 4, pp. 1195-1207, Jul. 2006
is because that the number of the subcarriers that may be [12] L.Dai, B. Gui and L.J.Cimini Jr., “Selective Relaying in OFDM Mul-
tihop Cooperative Networks,” in Proc. IEEE Wireless Communications
occupied by each relay node is decreased when K is large, and Networking Conference (WCNC ¡¯07), pp. 963¨C968, Hong Kong,
such that it is difficult to find the appropriate subcarriers for Mar. 2007.
each relay nodes. [13] I. Hammerstrom and A.Wittneben, “On the Optimal Power Allocation
for Nonregenerative OFDM Relay Links,” Proc. IEEE International Conf.
on Communications(ICC), Istanbul, Turkey, Jun. 2006.
VI. C ONCLUSIONS [14] I. Hammerstrom and A.Wittneben, “Power Allocation Schemes for
Amplify-and-Forward MIMO-OFDM Relay Links,” IEEE Trans. Wireless
In this paper, we have investigated the problem of dynamic Commun., vol. 6, no. 8, pp. 2798-2802, Aug. 2007.
subchannel and power allocation which has not been consid- [15] B. Gui and L.J.Cimini Jr., “Bit Loading Algorithm for Cooperative
ered previously in decode-and-forward cooperative OFDMA OFDM Systems,” EURASIP J. Wireless Commun. and Networking, vol.
2008, pp. 1-9, Jul. 2004.
systems. The proposed algorithm was designed to maximize [16] A. Han, et al., “Energy-Efficient Cooperative Transmission over Mul-
the minimal rate of the relay nodes under the constraints of tiuser OFDM Networks Who Helps Whom and How to Cooperate,” proc.
joint transmission power and subcarriers occupation. By for- of IEEE Wireless Commmunications and Networking Conference(WCNC),
vol. 2, pp. 1030-1035, New Orleans, Mar. 2005.
mulating the initial resource allocation problem to a standard [17] C. Muller, et al., “Dynamic Subcarrier Bit and Power Allocation in
convex optimization problem, the optimal solution is obtained. OFDMA-Based Relay Networks ,” in Proc. of 12th International OFDM-
However, the computational complexity of the optimal algo- Workshop, Hamburg, Germany, Aug. 2007.
[18] I. Maric and R. D. Yates, “Bandwidth and Power Allocation for
rithm is very high. To reduce the complexity without sacrific- Cooperative Strategies in Gaussian Relay Networks,” The 38th Asilomar
ing performance too much, we then designed two suboptimal Conference On Signals, Systems and Computers, Pacific Grove, pp. 1907-
algorithms. Simulation results indicated that the suboptimal 1911, CA, Nov. 2004
[19] S. Boyd and L. Vandenberghe, Convex Optimization. Cambridge, UK:
algorithm without SP could achieve the most computational Cambridge University Press, 2004.
complexity advantage but too much performance loss. At the
same time, the suboptimal algorithm with SP offers much com-