Cross-layer resource allocation approach in OFDMA ... - IEEE Xplore

Globecom 2013 - Communications QoS, Reliability and Modelling Symposium

Cross-Layer Resource Allocation Approach in OFDMA Systems with Multi-Class QoS Services and Users Queue Status Mustafa M Matalgah and Bimal Paudel

Omar M Hammouri

Department of Electrical Engineering University of Mississippi University, MS 38677–1848 Email: {mustafa, bpaudel1}@olemiss.edu

P.O. Box 234171 Encinitas, CA 92023 Email: [email protected]

Abstract—The design of an efficient and dynamic radio resource management (RRM) scheme is crucial for optimizing the resource utilization and providing QoS in wireless networks. Given the broad range of applications supported, high data rate required and low latency promised, the RRM scheme is indispensable for newly emerging air interface technologies such as WiMAX and LTE standards. The cross-layer information comes handy when judicious resource allocation and performance improvement are considered. This study proposes a crosslayer design for optimized resource allocation with quality of services (QoS) support that aims to balance between service provider’s revenue and subscriber’s satisfaction. The paper considers OFDMA system, which has been adopted in both WiMax and LTE, and presents a cross-layer design optimization for subchannel and power allocations with the objective of maximizing the capacity (in bits/sec/Hz) subject to fairness parameters and QoS requirement as constraints. The optimization does not only consider users’ channel conditions but also queue status of each user as well as different QoS requirements. The QoS classes adopted by the IEEE 802.16e standard, for WiMax technology, have been utilized in this study. In the proposed framework, the problem of power allocation is solved analytically while the subchannel allocation is solved using integer programming exhaustive search. The simulation and numerical results obtained in this paper have shown improved system performance as compared to other optimization schemes known in literature.

I.

I NTRODUCTION

Orthogonal Frequency Division Multiple Access (OFDMA) scheme has been adopted in the 4G WiMAX and LTE air interface technologies as their multiple-access mechanisms. Multiple access is achieved in OFDMA by assigning subsets of subcarriers and time slots to individual users. The subsets of subcarriers considered in frequency domain are referred to as subchannels. Radio resource management (RRM) involves mechanisms by which the system controls operations such as packet scheduling, admission control, subcarrier allocation, subchannel assignment, power allocation, modulation order, and rate control. The ultimate goal is to efficiently utilize the network resources and the scarcely available radio spectrum while keeping a good grade of services. A significant improvement in the performance of the wireless network can be realized by wisely adopting the cross-layer design approach for optimizing resource allocations in order to maximize a given

978-1-4799-1353-4/13/$31.00 ©2013 IEEE

1385

system utility [1]. Cross-layer design refers to protocol design done by actively exploiting the dependence between protocol layers to obtain certain performance gains. Non-cross-layer resource allocation algorithms operate only on the physical (PHY) layer. Some of the major noncross-layer techniques that have been reported in literature, which are based on maximizing the overall system throughput while having constraints on transmit-power are: maximum throughput approach, maximum fairness approach, proportional rate approach and proportional fairness approach [2][5]. In maximum throughput approach, the objective is to maximize the summation of rates from all users, while having a constraint on the total transmit-power [2]. The maximum fairness approach [3] focuses on achieving fairness among users while maximizing the minimum data rate among users. Therefore, it is called a max-min problem - maximizing the minimum data rate. The proportional rate constraints approach is a modification to maximum fairness approach, where the total throughput is maximized while maintaining some sort of proportionality among users’ data rates [4]. The proportional fairness aims to maximize the total throughput and minimize the total latency while maintaining fairness among users [5]. In most recent research studies reported in literature authors are focusing more on cross-layer design approaches [7]-[11]. The authors in [7] present a cross-layer scheme where the utility function, in terms of throughput, queue lengths and QoS requirements, is optimized subject to a frugality constraint dependent upon the queue information. In [8], a utility function subject to queue stability is optimized and the performance of the system was analyzed for various system parameters. In [9], a cross-layer resource allocation scheme for a system with imperfect channel-state information (CSI) was proposed. Delay offered by queue and outage probability offered by imperfect CSI were the constraints in the optimization problem. A better system performance was verified by simulation results. The study in [10] presents an algorithm to maximize the system throughput as a function of queue length subject to packet loss rate, average packet delay and average transmission rate. The effectiveness of the proposed scheme was verified by the simulation results. In [11] a cross-layer scheduling algorithm at the media access control (MAC) layer with multiple connections requiring diverse QoS requirement was proposed.

Globecom 2013 - Communications QoS, Reliability and Modelling Symposium

Given the literature review herein and to the best of our knowledge none of the work reported in literature addresses the problem of cross-layer optimization by taking into account the channel conditions, queue status and QoS requirements simultaneously. This paper addresses this issue and presents a resource allocation optimization scheme that takes into account both the channel conditions and the queue status of each user as well as different QoS requirements to maximize system throughput, which makes the proposed scheme unique to the state-of-the-art research on cross-layer optimization. Based on the queue status and QoS requirements of users, urgency and satisfaction factors are defined respectively, and are deemed as the cross-layer fairness parameters, which adds a new dimension to the fairness concept. Depending on the diverse QoS requirements of different users, resources can be allocated wisely; users that are well served and have no critical QoS requirements to schedule for service immediately can lag for some time allowing underserved users to access the channel. Optimization of the system performance subject to the constraints on power and subchannels as well as cross-layer fairness parameters are studied in this work. The significant improvement in the performance of the system in terms of maximization of system throughput achieved with the implementation of the proposed cross-layer approach is justified by the extensive simulation results. The remainder of the paper is organized as follows. Section II presents the proposed cross-layer design approach for maximizing system capacity (in bits/sec/Hz) with system resources, QoS requirements, and users queue lengths as constraints. This section illustrates the systems model, problem formulation, and proposed solutions. The simulation and numerical results using the proposed approach and comparisons with other techniques known in literature are provided in Section III. Finally, some conclusions are drawn in Section IV.

II.

P ROPOSED C ROSS - LAYER S CHEME

Resource allocation is concerned with the proportional allocation of resources such that a system utility is optimized. Allocating resources to users based on the channel conditions of each user is one of the approaches for resource allocation and system throughput optimization. The other approach of throughput optimization is the cross-layer approach where queue status of each user as well as different QoS requirements are considered along with the channel conditions. In order to efficiently utilize the limited sub-channels and power in an OFDMA system, the buffer condition of each user should be taken into consideration. Cross-layer design has received attention most recently in literature and used to achieve multiuser diversity gain where information about the channel state at the physical layer are passed on to the packet schedulers at the MAC layer [5], [12]. The merits of passing queue length and QoS information from the MAC layer to the physical layer have been studied in the literature. In [13], it has been proved that in order to minimize the total number of packets in the system or to maximize the system throughput, resources should be allocated to the user with not only the best channel conditions but also the longest queue.

1386

A. System model A downlink OFDMA system is considered, where the base station (BS) serves M users with a total of K connections from all users. Further, the total bandwidth, B, is divided into L subchannels. Hence, the bandwidth of each subchannel is B/L and the time slot duration corresponding to each subchannel is L . Users can have multiple connections at certain time. Ts = B Data from these connections arrive from the MAC layer and is placed into a infinite buffer depending on the service flow associated and the corresponding user. These buffers follow a first in first out (FIFO) strategy. A channel fading that follows rayleigh distribution with envelope hk,l is assumed to be experienced by connection k over subcarrier l. Moreover, the following assumptions are made: i) outgoing queues for every connection are infinite; ii) the BS has perfectly received the channel state information (CSI) from all subscriber sets (SSs); iii) the subchannel and power allocation information is sent to each user on a separate channel; iv) coherent bandwidth of the channel is larger than B L , that is, the channel response on each subchannel is flat; v) the channel gain remains fixed during each time slot Ts ; and vi) the channel is varying in time slow enough that users can estimate the channel CSI perfectly. B. Quality of Service (QoS) parameters IEEE 802.16e defines different service classes: Unsolicited Grant Service (UGS), Real Time Polling Service (rtPS), Extended Real Time Polling Service (ErtPS), Non Real Time Polling Service (nrtPS) and Best Effort (BE) and also defines certain QoS parameters associated with each of these classes. Different service classes support different applications with some defined QoS parameters [14], [15]. Two cross-layer parameters are introduced in this section that can be deemed as QoS parameters: 1) Service Urgency: Service urgency proposed here is a cross-layer QoS parameter that is dependent on the information about the queues of the services in the data link layer. Let Ak (n) be the number of bits arriving at connection k’s queue during frame n, Qk (n) be the length of queue associated with connection k during frame n and Bk (n) be the number of bits the BS serves from the queue of connection k during frame n. Then the queue length of connection k during frame n + 1 is given by Qk (n + 1) = Qk (n) + Ak (n) − Bk (n). Every connection can be associated with one of the five different service flows. Let SF x denote an xth service class where x is any element in the set {UGS, ErtPS, rtPS, nrtPS, BE} and SF xx(k) denote SF x associated with connection k. Also, let with the same ΩSF be the set of all connections associated x SF x . Then this set is expressed as ΩSF (k) = {1 ≤ k ≤ K : SF x (k) = xSF x }∀x ∈ {U GS, ErtP S, rtP S, nrtP S, BE}, and let QSF (n) be the aggregate queue length of connections associated with the same service class during frame n, then x QSF (n) can be expressed as  x QSF (n) = Qk (n). (1) kQSF x

Finally, the normalized queue length of connection k during frame n, Uk (n), which will be called henceforth the Urgency Factor, can be defined as  Qk (n) , SF x ∈ {rtPS, nrtPS, BE} QSF x (n) (2) Uk (n) = 1, SF x ∈ {UGS, ErtPS}.

Globecom 2013 - Communications QoS, Reliability and Modelling Symposium

It should be noted here that the Urgency Factor Uk (n), is set to 1 for connections with a UGS or ErtPS service flow type. It is known from the QoS requirements that the connections associated with UGS and ErtPS service classes should be allocated resources on a periodic basis and therefore the concept of urgency does not apply. It should also be noted that since Uk (n) ∈ [0, 1], UGS and ErtPS service flows are assigned the highest urgency factor of 1. However, the urgency factor for rtPS, nrtPS and BE is calculated using (2). Here the queue length for a given connection k at a given frame number n corresponding to a given service flow type SF x is normalized by the total queue length of that particular SF x in the system. This normalized urgency factor will be high for the service type with the longest queue at a given connection k at a given frame number n and hence this connection will be served first. The significance of the Urgency Factor is twofold. It gives indication about which connection is being underserved relative to other connections of the same service flow, and it also conveys information about the queue length of the connection to the resource-allocation algorithm. The higher the value of Uk (n), the more it is urgent to allocate resources to the connection. 2) Service Satisfaction: Service satisfaction based on different kinds of service flows depends on the information like data rate, delay satisfaction indicator or flow’s coefficient as defined in [11]. Hence, service satisfaction can be deemed as a cross-layer QoS and is considered in this study. Let {γU GS , γErtP S , γrtP S , γnrtP S , γBE } be defined as a set of configurable system parameters. Each γSF denotes a weighting factor that can be used to favor one service class over the other and be configurable depending on the system deployment. For example, if the priority order for different QoS classes is UGS>ErtPS>rtPS>nrtPS>BE, then the weighting factors can be set under the constraint γU GS > γErtP S > γrtP S > γnrtP S > γBE , e.g., γU GS = 10 > γErtP S = 9 > γrtP S = 7 > γnrtP S = 6 > γBE = 2. Now let Sk (n) be the satisfaction factor for connection k during frame n. For UGS connections the satisfaction factor is defined as 1 , (3) Sk (n) = γU GS where γU GS is the UGS class weighting factor. Therefore, the satisfaction factor is constant for all UGS connections and associated frames. As for ErtPS connections, the satisfaction factor is defined as 1 , (4) Sk (n) = γErtP S where γErtP S is the ErtPS class weighting factor. Also, the satisfaction factor is constant for all ErtPS connections and associated frames. ErtPS connections generate constant size packets like UGS but, unlike UGS, packets are not generated periodically. Therefore, whenever data is available it is treated the same as UGS data. For rtPS connections, the satisfaction factor is defined as DSk (n) , (5) Sk (n) = γrtP S where γrtP S is the rtPS class weighting factor and DSk (n) is the delay satisfaction indicator, which is defined as DSk (n) = max{1, Tk − ΔT − Wk (n) + 1},

(6)

1387

where Tk is the maximum latency for connection k, ΔT ∈ [0, Tk ] is the guard time region ahead of the deadline Tk , and Wk (n) ∈ [0, Tk ] is the head of line (HOL) delay. Let’s consider the two extreme cases here. First, if HOL delay and guard time for a connection, Wk (n) and ΔT both ∈ [0, Tk ] are 0 then the delay satisfaction indicator will be 1 + Tk , maximum attainable value from (6), indicating that the connection is served well and no more resource need to be allocated for that connection. Second, if HOL delay and guard time for a connection, Wk (n) and ΔT both ∈ [0, Tk ] are Tk then the delay satisfaction indicator will be 1, minimum attainable value from (6), indicating that the connection is underserved and more resources should be allocated to meet the maximum delay requirements. The satisfaction factor for rtPS connections 1 . For nrtPS connections, the has a minimum value of γrtP S satisfaction factor is defined as RSk (n) Sk (n) = , (7) γnrtP S where γnrtP S is the nrtPS class weighting factor, RSk (n) is the rate satisfaction indicator, which is defined as RSk (n) = max{1, ηˆk (n)/ηk },

(8)

where ηk is the minimum reserved data rate for connection k, and ηˆk (n) is the exponentially weighted average data rate for connection k up to frame n obtained by using the exponentially weighted low-pass filter [6]. Exponential weighted average or the exponential smoothing is performed on the time series data such that the data can be obtained in smooth and presentable form. In traditional weighted averaging, equal weights are assigned to the past observations however, in the exponential smoothing decreasing exponential weights are assigned to time series data. This exponential weighted averaging based on the exponential low-pass filter as applied to the data rate calculation over number of frames can be defined as  Ck (n), n=0 (9) ηˆk (n + 1) = ηˆ (n)(1 − 1 ) + C (n) 1 , n > 0, k

tc

k

tc

where Ck (n) is the transmit rate allocated during frame n to connection k. The parameter tc , window size, controls the latency of the system. If tc is large, then the latency increases, with the benefit of higher throughput. If tc is small, the latency decreases since the average throughput values change more quickly, at the expense of some throughput [15, pp. 213]. The satisfaction factor, Sk (n), ensures that the connection is receiving an average transmission rate above the minimum reserved rate, ηˆk (n) ≥ ηk . If RSk (n) ≥ 1, then the rate requirement is satisfied, which increases the satisfaction factor. Large values of RSk (n), therefore, indicate high degree of satisfaction. The minimum value for RSk (n) is 1, which is when the connection is underserved and should be allocated more resources to meet the minimum rate requirements. The satisfaction factor for nrtPS connections has a minimum value 1 . For BE connections, the satisfaction factor is defined of γnrtP S as 1 , (10) Sk (n) = γBE where γBE is the BE class weighting factor. Therefore, the satisfaction factor is constant for all BE connections and associated frames. The reason is that by definition of the QoS requirements, BE connections should be allocated resources

Globecom 2013 - Communications QoS, Reliability and Modelling Symposium

after all other service flows are satisfied, and therefore, the concept of service satisfaction does not apply. The significance of the satisfaction factor is also two-fold. It allows for scalability, as when the system is overloaded, the performance of connections with low-priority service classes will be degraded prior to those with high priority service classes, and it also allows connections with low-priority service classes to lead when connections with higher-priority service classes are well satisfied.

the total available power. The second constraint states that the power for all subcarriers should be positive or zero. In the third constrain, ρk,n is only allowed to be 0 or 1 which means a user is not allowed to use a portion of a subcarrier. Furthermore, no sharing of subcarriers is allowed, which is stated by the fourth constraint. The last constraint is the fairness constraint presented in (12).

C. Problem Formulation

The optimization problem given in (13) is very hard to solve. It is a mixed binary integer programming problem. The problem has nonlinear constraints as well as continuous variables, Pk,l , and binary variables, ρk,l . An optimum solution to this problem is highly computationally complex, so a suboptimal solution is proposed, where the optimality is compromised for complexity reduction. An analytical solution to the optimization problem in (13) is obtained by adopting the approach presented in [4]. The problem in (13) can be reformulated for a given subchannels allocation and defined by a cost function. Lagrangian method of optimization problem solution is then used. This method results in an expression for H −Hk,1 . optimal power allocation as Pk,l = Pk,1 + Hk,n k,n Hk,1

Let Pk,l (n) be the power allocated to connection k over subcarrier l during frame n, σ 2 be the variance of additive white Gaussian noise (AWGN) random variable (assumed to have zero mean), hk,l be the channel gain for connection k on subcarrier l, and ρk,l ∈ {0, 1} be an indicator that indicates whether a subcarrier l is used by user k or not. Then, the transmit rate allocated to connection k during frame n is expressed L ρk,l as Ck (n) = l=1 L log2 [1 + Pk,l (n)Hk,l (n)], bps/Hz where Hk,l (n) = h2k,l /σ 2 B L , and the weighted transmit rate, Rk (n), is Uk (n) Rk (n) = (11) Ck (n). Sk (n) The weighted transmit rate in (11) incorporates both the urgency factor and the satisfaction factor. It is known from the explanation of urgency factor that the services with higher queue lengths have higher urgency and should be scheduled first. So, higher service urgency requires a higher transmit rate to be allocated to that service hence transmit rate is directly proportional to the urgency factor. However, the satisfaction factor indicates the satisfaction level of the service. If a service meets a specified data rate requirement, a delay requirement or any other requirements specific to the QoS, then the satisfaction is high such that a decrease in the transmit rate could still meet the QoS requirements of that service. So, the transmit rate is inversely proportional to the service satisfaction. Accordingly, the resultant transmit rate allocated to the connection increases with urgency and decreases with more satisfaction. Now the fairness constraint taking into account both the urgency and satisfaction of the service flows is defined as Ri (n) = Rj (n) = R(n)

∀i, j ∈ [1, K].

(12)

Based on these discussions, the optimization problem can be expressed mathematically as max

Pk,l ,ρk,l

subject to

Ck =

K  L  ρk,l

k=1 l=1 K  L 

L

log2 (1 + Pk,l Hk,l ) (13)

Pk,l ≤ Ptot

(14)

k=1 l=1

Pk,l ≥ 0 ∀ k, l ρk,l = {0, 1} ∀ k, l K  ρk,l = 1 ∀ l

(15) (16) (17)

k=1

Ri (n) = Rj (n) = R(n) ∀ i, j ∈ [1, K],(18)

where Ptot is the total available power. The first constraint implies that the total power over all subcarriers is not to exceed

1388

D. Problem Solution: Two-phase greedy approach

A two-phase greedy approach, that results in a suboptimal solution of the optimization problem is formulated in this study. The two-phase greedy approach starts off with allocating equal power to all the subchannels. Later, power is allocated in order to maximize the total system throughput while maintaining fairness and QoS support. This approach is both the mixture of analytical solution and integer programming exhaustive search. The proposed resource allocator (RA) algorithm based on the two-phase greedy approach is shown in Algorithm 1. The terms used in this algorithm are defined as follows: T is the total traffic duration, Tc & Ts are the in-phase & quadrature phase E-field components of Rayleigh fading channel respecy is the initial tively, hk,l and ρk,l are as defined above Pinit power allocation to y resource allocation approach, where y is either Proposed (prop), Proportional Rate Constraint (PRC) Maximum Fairness (MF) or Maximum Sum Rate (MSR) method, Cky is the throughput corresponding to k th user and y method and C¯ky is the exponentially weighted average capacity corresponding to k th user and y method. The working of the resource allocator algorithm depicted in Algorithm 1 is described in detail in the following. It first reads the queues lengths, {Qk (n)}, connections service flows, SF x (k), the maximum delay accepted for every rtPS connection, Tk , the minimum data rate accepted for every nrtPS, ηk (n), and the HOL delay for every rtPS connection, Wk (n), from the MAC layer. Likewise, total power, Ptot , total available bandwidth, B, total number of subcarriers, L, the noise variance, σ 2 , service flow weighting factors, {γSF }, window size, tc , and the guard time ahead of a deadline, ΔT are configured by the allocator. With all the information in hand traffic corresponding to different types of QoS classes are simulated based on the parameters listed in Table I. Table I summarizes the characteristics of 10 different traffic in the system based on five different QoS classes. Queue lengths corresponding to a particular service flow are then calculated. Service urgency and satisfaction factors are then evaluated

Globecom 2013 - Communications QoS, Reliability and Modelling Symposium

and a Rayleigh fading channel based on Clarks’s model is simulated. The algorithm then proceeds forward with the procedure for subchannel allocation, which starts with finding the subchannel with maximum channel gain for each user in Rayleigh fading environment. This subchannel with maximum gain is then assigned to the user. The user data rate supported for that particular subchannel is then calculated. Likewise, a data rate that compensates for the fairness in resource allocation is calculated using the weighting factor for all the users. The user with the minimum fairness compensated data rate Rk is then found. Since the main purpose here is to maximize the capacity of the system, user supporting minimum Rk found in preceding step is given first priority and yet another subchannel is allocated to that user such that the total data rate for that user is increased. So, a subchannel with maximum gain is allocated to the user supporting minimum Rk . This process is repeated until all the subchannels are allocated to the users. Once the subchannel allocation is completed, power allocation as explained in [4] is performed. Finally the overall system capacity is evaluated. TABLE I: Traffic Simulation Parameters Connection

SF

1

UGS

2

UGS

3

rtPS

4

rtPS

5

ErtPS

6

ErtPS

7

nrtPS

8

nrtPS

9

BE

10

BE

III.

Parameter

Value

CODEC VPI Packet suze CODEC VPI Packet size Bernoulli trial (p) ¯ Mean rate Maximum delay Bernoulli trial (p) ¯ Mean rate Maximum delay CODEC VPI Packet size Mean ON period Mean OFF period CODEC VPI Packet size Mean ON period Mean OFF period Mean rate Minimum rate Mean rate Minimum rate Pareto mean rate Lognormal mean rate Pareto mean rate Lognormal mean rate

G.729 20 ms 66 Bytes G.728 30 ms 106 Bytes 0.4 64 Kbps 30 ms 0.5 16 Kbps 50 ms G.723.1 30 ms 66 Bytes 1.2 sec 1.8 sec G.711 20 ms 206 Bytes 1.2 sec 1.8 sec 512 Kbps 128 Kbps 1 Mbps 1 Mbps 10558 bps 7247 bps 10558 bps 7247 bps

Algorithm 1: Resource Allocator Input: Ptot , L, K, B, N, σ 2 , γSF x , ΔT, tc , Tk , Wk (n), ηˆk (n) Initialize Array: C¯k = 0, QSF = 0 for t = 1 → T do generate QSF x ∀ x ∈ {U GS, rtP S, ErtP S, nrtP S, BE} end for n = 1 → N do x f ind QSF (n) // by (1) // by (2) f ind Uk (n) // by (3)-(10) f ind Sk (n) generate Tc & Ts hk,l ⇐ Tc + jTs assign ρk,l // subchannel allocation for k = 1 → K do  prop tot Pinit (k) ⇐ PL ρk,i i

P RC MF Pinit (k) & Pinit (k) // as in [4] y evaluate : Ck ∀ y ∈ {prop, P RC, M F } //by (??)-(11) end K  Cky ∀ y ∈ {prop, P RC, M F } C y (n) = k=1

if n = 1 then C¯ y (n) = C y (n) ∀ y ∈ {prop, P RC, M F } else C¯ y = C¯ y (n − 1) ∗ (1 − t1c ) + C y (n) ∗ t1c ∀ y ∈ {prop, P RC, M F } end end

TABLE II: Simulated System Parameters

S IMULATION AND R ESULTS

In this section we numerically implement and simulate the solution described in Section II-D for the optimization problem presented in Section II-C based on Algorithm 1. Table II shows the system parameters used in this simulation. The values listed in the table are not changed across the different scenarios assumed in this simulation. Table I lists various parameters corresponding to different kinds of traffic assumed in the system and the traffic simulation is based on these listed parameter. The wireless channel model used here is same as in [4].

1389

Symbol

Parameter

Value

Ptot L K N B σ2 Tf ΔT Tc γU GS γrtP S γErtP S γnrtP S γBE

Total system power Number of subchannels Number of connections Number of frames Total system bandwidth AWGN variance Frame duration Guard time ahead of deadline Moving average window size UGS weighting factor rtPS weighting factor ErtPS weighting factor nrtPS weighting factor BE weighting factor

1 Watt 64 10 1000 5 MHz 1.1565 × 10−8 5 ms 20 ms 1000 ms 0.8 0.6 0.4 0.3 0.2

The results in Fig. 1 show the total average system capacity versus average SNR based on the simulation parameters as listed in Table II. The figure compares the performance of different resource allocation schemes along with the proposed cross-layer scheme. The figure also shows the Shannon’s theoretical limit. It can be seen from the figure that the proposed algorithm outperforms other schemes in terms of average capacity throughout the whole SNR range (-5 to 30 dB) and

Globecom 2013 - Communications QoS, Reliability and Modelling Symposium

Shannon’s limit than the other optimization schemes. Similarly it has also been observed that the proposed scheme maintained the optimum capacity of the system throughout the whole frames considered. These simulation results provide a clear justification that the implementation of the proposed crosslayer resource allocator scheme results in improved system capacity performance. While IEEE 802.16e (WiMAX) QoS classes have been utilized in developing the work in this paper, the work can also be extended to LTE standard QoS classes where a common QoS treatment is offered to service data flows mapped to the same bearer.

Total Average System Capacity (bps/Hz)

12

10

8

6

4 Shannon Limit Proposed Proportional Rate Constraint [4] Maximum Fairness [3]

2

0 −5

0

5

10 15 Eb/N0 (dB)

20

25

R EFERENCES

30

Fig. 1: Total average system capacity vs. average SNR

Total Average System Capacity (bps/Hz)

7 6

5

4

3

2 Proposed Proportional Rate Constraint [4] Maximum Fairness [3]

1

0

0

200

400 600 Frame Number

800

1000

Fig. 2: Total average system capacity vs. frame number

is close to the Shannon’s limit. Likewise, the simulation result in Fig. 2 depicts the total average system capacity versus frame number. It can be seen from the graph that the average capacity of the system remains at the optimum and constant level for an extended period of traffic time (1000 frames for the case considered). It can also be seen that the proposed system outperforms the other resource allocation schemes. IV.

C ONCLUSION

In this paper a resource allocation optimization scheme for OFDMA systems with multi-class QoS and user queue status is presented. The scheme is designed to support QoS and fairness among users while maximizing the system capacity. A cross-layer approach is followed to share information between the MAC and the PHY layers. Furthermore, the proposed scheme is simulated and a comparison has been made between the performance of the scheme and major techniques known in literature. It has been observed that the proposed scheme resulted in total average system capacity that is closer to the

1390

[1] S. Shakkottai, T. Rappaport, and P. Karlsson, “Cross-Layer Design for Wireless Networks,” IEEE Communications Magazine, vol. 41, no. 10, pp. 74-80, 2003. [2] Y. Zhang and K. Letaief, “Multiuser adaptive subcarrier-and-bit allocation with adaptive cell selection for OFDM systems,” IEEE Transaction on Wireless Communications, vol. 3, no. 5, pp. 1566-1575, 2004. [3] W. Rhee and J. Cioffi, “Increase in capacity of multiuser OFDM system using dynamic subchannel allocation,” VTC 2000-Spring Tokyo. 2000 IEEE 51st Vehicular Technology Conference Proceedings, 2000, vol. 2, 2000. [4] Z. Shen, J. Andrews and B. Evans, “Adaptive resource allocation in multiuser OFDM systems with proportional rate constraints,” IEEE Transactions on Wireless Communications, vol. 4, no.6, pp. 2726-2737, 2005. [5] D. Tse, “Forward Link Multiuser Diversity Through Proportional Fair Scheduling,” Bell Laboratories Journal, 1999. [6] P. Vuswabatg, D. Tse and R. Laroia, “Opportunistic beamforming using dumb antennas,” IEEE Transactions on Information Theory, vol. 48, no. 6, pp. 1277-1294, 2002. [7] B. Danobeitia and G. Femenias, “Cross-Layer Scheduling and Resource Allocation in OFDMA Wireless Networks,” in Wireless Days (WD). 2011 IFIP, pp. 1-6, 2011. [8] A. G. Marques, L. M. Lopez-Ramos, G. B. Giannakis, J. Ramos and A. J. Caamano, “Optimal Cross-Layer Resource Allocation in Cellular Networks Using Channel- and Queue-State Information,” IEEE Transactions on Vehicular Technology, vol. 61, no. 6, pp. 2789-2807, 2012 [9] N. Mokari, M. R. Javan and K. Navaie, “Cross-Layer Resource Allocation in OFDMA Systems for Heterogenous Traffic With Imperfect CSI,” IEEE Transactions on Vehicular Technology, vol. 59, no. 2, pp. 1011-1017, 2010. [10] H. Tian, H. Xu, Y. Gao, S. Wang and P. Zhang, “QoS-Oriented Crosslayer Resource Allocation with Finite Queue in OFDMA Systems,” in Wireless Communications and Networking Conference, WCNC 2008, pp. 1519-1524, 2008. [11] Q. Liu, X. Wang and G. B. Giannakis, “A Cross-Layer Scheduling Algorithm With QoS Support in Wireless Networks,” IEEE Transactions on Vehicular Technology, vol. 55, no. 3, pp. 839-847, 2006. [12] R. Knopp and P. Humblet, “Information Capacity and Power Control in Single Cell Multiuser Communications,” in ICC 95 Seattle, Gateway to Globalization. IEEE International Conference on Communications, vol. 1, pp. 321-335, 1995. [13] L. Tassiulas and A. Ephremides, “Dynamic Server Allocation to Parallel Queues with Randomly Varying Connectivity,” IEEE Transactions on Information Theroy, vol. 39, no. 2, pp. 466-478, 1993. [14] “IEEE Standard for Local and Metropolitan Area Networks - Part 16: Air Interface for Fixed Broadband Wireless Access Systems, Amendment 2: Physical and Medium Access Control Layers for Combined Fixed and Mobile Operation in Licensed Bands and Corrigendum 1,” IEEE Std. 802.16e-2005 and IEEE Std 802.16-2004/Cor1-2005, 2005. [15] J. Andrews, A. Ghosh, R. Muhamed, Fundamentals of Wimax: Understanding Broadband Wireless Networking. Prentice-Hall PTR Upper Saddle River, NJ, USA, 2007.