Joint channel allocation and power control with fairness in multi-hop cognitive radio networks①

Pei Xuebing (裴雪兵)

(*China Ship Development and Design Center, Wuhan 430064, P.R.China)(**School of Computer and Electronics Information, Guangxi University, Nanning 530004, P.R.China)



Joint channel allocation and power control with fairness in multi-hop cognitive radio networks①

Pei Xuebing (裴雪兵)②

(*China Ship Development and Design Center, Wuhan 430064, P.R.China)(**School of Computer and Electronics Information, Guangxi University, Nanning 530004, P.R.China)

This paper investigates channel allocation and power control schemes in OFDM-based multi-hop cognitive radio networks. The color-sensitive graph coloring (CSGC) model is viewed as an efficient solution to the spectrum assignment problem. The model is extended to combine with the power control strategy to avoid interference among secondary users and adapt dynamic topology. The optimization problem is formulated encompassing the channel allocation and power control with the interference constrained below a tolerable limit. Meanwhile, the proposed resource allocation scheme takes the fairness of secondary users into account in obtaining the solution of optimization. Numerical results show that the proposed strategy outperforms the existing spectrum assignment algorithms on the performance of both the network throughput and minimum route bandwidth of all routes, as well as the number of connected multi-hop routes which implies the fairness among secondary users.

channel allocation, power control, multi-hop cognitive radio networks

0 Introduction

The spectrum usage is concentrated on certain portions of the spectrum while the assigned spectrum ranging from 15% to 85% remains unutilized according to Federal Communications Commission (FCC). The limited available spectrum and the inefficiency in spectrum usage necessitate a new communication paradigm to exploit the existing wireless spectrum opportunistically[1]. The cognitive radio techniques enable the usage of temporally unused spectra, which are referred as spectrum holes or white space. NeXt Generation (XG) communication networks, also known as Dynamic Spectrum Access Networks (DSAN) as well as Cognitive Radio networks, can provide high bandwidth to mobile users through dynamic spectrum access techniques in heterogeneous wired/wireless networks environment integrating cellulars, WLAN, WiMAX, MANET, PSTN networks, etc.

In Ref.[2], a mathematical model was proposed to reduce the spectrum allocation problem to color-sensitive graph coloring (CSGC) under a fixed topology, and described a set of centralized and distributed approximation algorithms to optimize the different reward functions. However, power control is not considered in Ref.[2], and the protection for primary users is simplifized as the physical separation. The disadvantage of this distance-based interference model is that it does not take into account the aggregated interference effect when multiple transmissions happen on a single channel.

In Ref.[3], the authors studied the problems of secondary spectrum access with minimum SINR guarantee and interference temperature constraint. However, the solution to the social spectrum optimization may be highly complex and infeasible when all of secondary links are active. It is also noted that, Ref.[3] focused on single-channel scenario and channel allocation was not of concern. Ref.[4] presented an admission control algorithm jointly with power control with explicit interference protection for primary users and QoS guarantees for secondary users in heavy load condition in centralized cognitive CDMA wireless networks. Obviously, it was also for single-channel scenario and could not be applicable for the case of multi-hop cognitive radio networks.

Ref.[5] considered the centralized downlink channel allocation and power control problem to maximize the number of users in single-hop cognitive radio networks. In Ref.[6], the author discussed both downlink and uplink resource allocation in a distributed manner to maximize the throughput of a cognitive radio network with cooperation between primary users and secondary users while maintaining the performance of coexistent primary users. However, these optimizations focus on the case of single hop network. While in this paper, the case of multi-hop cognitive radio network is considered due to the need of broadening the coverage of cognitive radio base station.

Our paper is also related to Refs[7] and [8], where a problem of joint channel assignment and transmission power control for multi-channel wireless ad hoc networks is considered. In Ref.[9], joint link scheduling and power control were researched with the objective of maximizing network throughput in a TDMA-based multi-hop wireless networks, but it is not suitable for OFDM-based multi-channel cognitive radio networks. In a broader context, the conventional problem of power control is also researched in cellular networks. As the work is not for cognitive radio network, protecting primary users is not of concern.

In this paper, the channel allocation model and power control are investigated in cognitive radio networks, while most work on OFDM-based channel allocation is based on fully-connected single hop wireless networks. In order to improve frequency efficiency and optimize the system revenue, a joint bandwidth allocation and power control problem is formulated with interference protection for primary users.

The rest of this paper is organized as follows. Section 1 describes the centralized system model of cognitive radio networks. The optimization of joint channel assignment and power control is formulated in Section 2 and the proposed heuristic approach will also be demonstrated specifically in Section 2. The numerical results are presented in Section 3. Finally, Section 4 draws a conclusion on this paper.

1 System model

For exploiting the licensed underutilized spectrum by opportunistic access, mobile users can be formed into secondary multi-hop cognitive radio networks. They individually or cooperatively sense unoccupied spectrum holes and communicate with cognitive radio base stations through multi-hop relaying. The system model is depicted in Fig.1. The spectrum of interest is divided into K channels that are licensed to a primary network of M PTX-PRX links. Each primary link occupies one of the K channels. Furthermore, each channel can be used by one or several secondary users by controlling the transmission power for the purpose of dominating the interference to the primary users. The cognitive radio network consists of one base station serving a set of S nodes with the capability of cognitive radio.

Fig.1 Deployment of a cognitive radio network

Assume that all secondary mobile users communicate with cognitive radio base station through a common control channel to exchange control information, for example the bandwidth allocation and transmitting power level, and so on. Moreover, source node can always build an optimum multi-hop route to the cognitive radio base station, the cross-layer challenges about route selection and spectrum management will be discussed in our future work. And the orthogonal frequency division multiplexing (OFDM) is adopted in the multi-hop cognitive radio networks, because OFDM technique has the advantage of feeding certain subcarriers with zero, thus exploits the available but non-contiguous wireless spectrum. The frequency separation among subcarriers can be common or different.

2 Joint channel allocation and power control

2.1 Problem definition

(1)

(2)

(3)

Using adaptive coding and/or modulation, the transmission rate of secondary user can be varied according to the received SINR. It can be written as Ref.[6]

(4)

It is assumed that, there are S secondary users to compete K channels in multi-hop cognitive radio networks. They are indexed from one to, S and K respectively. Let ai,cbe a binary variable that indicates whether channel c is assigned to secondary user i. In particular, ai,cis set to 1 if the channel is assigned to user i. Otherwise, ai,cis set to 0. Then, the link throughput of the l’th link of the n’th route can be calculated as

(5)

where K is the number of channels and i is the transmission node of the l’th link.

Suppose that there are N multi-hop routes to BS and the number of links in n’th route is L(n). In other words, N is the number of secondary users with source services. Therefore, for maximizing the throughput of secondary multi-hop cognitive radio networks, the optimization model can be formulated as

(6)

(7)

2.2 Solution to the optimization model with fairness

The optimal channel assignment combined with power control is known to be NP-hard. Therefore, the solutions of optimization problems Eq.(6) can only be solved through heuristic based approaches according to the following procedures.

For the purpose of considering the fairness among secondary users, the process of allocating channels will give higher priority to the secondary user with the least idle channels and number of hops.

The resource allocation algorithm can be described as follows.

Step 1: determining the priority of secondary links for allocating channels to the secondary users. We first need to determine the priority of all secondary routes based on the number of hops. The smaller the number of hops, the higher the priority. When the number of hops in the secondary routes is the same, the priority can be determined by the number of idle channels. The less the number of idle channels, the higher the priority of the secondary route.

When one route is allocated to the idle channels according to the given priority, all of the links in the route should be allocated at least one idle channel in order to make the route usable. After all of links in the route is allocated the idle channels, another route with the lower priority is allocated to the remaining idle channels until all the idle channels are allocated completely.

In order to insure the fairness among the secondary routes, all links of each route can be allocated to only one idle channel at one computation loop. After one computation loop is completed, the priority of all secondary routes is analyzed again according to the number of hops and remaining idle channels. The allocation of channel and power is carried out again according to the new priority.

Step 2: The transmission power should be allocated for each link after it is allocated idle channel. When each ai,c(with channel index c=1,2,…,K for each secondary user i with i=1,2,…,S) is determined, it is needed to compute the transmission power for each link of secondary users. Firstly, initializing the transmission power of secondary users, the initial power can be determined by the following formula

(8)

After one link is allocated to the transmission power, the next turn is the next link of the same secondary route. The next link is allocated to the idle channel and transmission power according to the same method.

Step 3: Termination

When all links of all secondary routes are allocated to one idle channel and power, one computation loop is completed. Then, the next loop begins again through computing the new priority of all routes. The computation loop does not terminate until there is no idle channel.

When there is no idle channel, the solution to the optimization model is obtained with the consideration of fairness among the secondary routes.

3 Simulation results and discussion

In this section, some numerical simulations are conducted to evaluate the performance difference between our proposed bandwidth allocation jointly with power control strategy and traditional algorithms. The referenced traditional algorithms are Collaborative-Max-Sum-Reward (denoted as CSUM) and Collaborative-Max-Min-Reward (denoted as CMIN) algorithms proposed in Ref.[2]. Obviously, they are link-oriented multi-hop spectrum assignment algorithms and do not take into account the factor of multi-route some nodes support. The power control technique among secondary users is also not adopted in the referenced algorithms. Thereby, the interference constraints in both of CSUM and CMIN approaches are obtained by comparing the distance relationship between secondary users.

3.1 Simulation configuration

A deterministic 20-nodes network topology is adopted defined in Ref.[10] to investigate the performance. What’s more, the square area with 20-nodes network topology is expanded to 2000×2000 centered by origin. Therefore, all of the coordinates of 20 nodes in Ref.[10] are multiplied by 40 and then subtracted by 1000 consistently. Five secondary users and three primary users with QoS services are selected randomly among the 20 nodes, and the other secondary users act as the relaying nodes of multi-hop cognitive radio network. The secondary users access the IP core network through cognitive radio base station located in origin.

Suppose that channel gain G for primary and secondary links can both be formulated as

G=d-ρ

(9)

where d is the physical distance between transmitter and receiver, and ρ is the path attenuation exponent.

The MATLAB is used as the simulator to perform the recurrent search of channel assignment and power calculation. The optimization is carried out by the central infrastructure such as cognitive radio base station, and the other secondary users only implement the multi-hop relay function. Some common simulation parameters are listed in Table 1.

Table 1 Simulation parameters

3.2 Simulation results and discussions

In the simulation scenery, the total number of channels is changed in each simulation and the unoccupied channels detected by each node are produced at random.

Fig.2 shows that the network throughput of all secondary routes achieved from the proposed channel and power allocation strategy is larger than that of both CSUM and CMIN allocation algorithms when the total number of channels changes. This is because the proposed allocation approach can make the best use of channel resources and maximize the network throughput by allocating channels to all of the links of one route symmetrically. Thus, link bandwidth is easier to contribute to the route bandwidth for symmetry. However, the optimization of CSUM and CMIN approaches focusing on the links cannot efficiently achieve the bandwidth revenue of one route for the unbalancing of channel allocation, which will make the resource allocation ineffective. On the other hand, the power control technique can efficiently increase the channel reuse and achieve the gain of route bandwidth.

Fig.2 Network throughput varying with the number of channels

Fig.3 validates the efficiency of the proposed channel and power allocation strategy better than both CSUM and CMIN allocation algorithms for the optimization of minimum bandwidth among the secondary routes. It is clear that our proposed algorithm considers the fairness among the secondary users and allocates channels to the multi-hop links according to the priority of secondary routes in turn. Therefore, our proposed optimization has the advantage of fairness among secondary routes because the bandwidth and power assignment always favors the route with the weakest channel condition, especially when idle channels lack for secondary users.

However, the CSUM and CMIN optimization approaches are easy to engender the starvation phenomenon. As we expected, there are always secondary routes getting no opportunity of data transmission according to both CSUM and CMIN algorithms which can be seen from the simulation results in Fig.3. There are two reasons to account for this starvation phenomenon. Firstly, both CSUM and CMIN optimizations focus on channel allocation for link and can not make an efficient contribution to the bandwidth of secondary routes. Secondly, serious conflict will happen when the power control technique is not adopted. Due to finite channels and rigorous interference constraints for the primary users, there exists a certain link of one route obtaining no idle channels.

Fig.3 The minimum bandwidth of secondary routes varying with the number of channels

The optimization results for the number of connective multi-hop routes are also analyzed as in Eq.(7). Simulation results in Fig.4 demonstrate the fairness of the proposed channel and power allocation strategy among secondary users from the viewpoint of number of connective multi-hop routes. It can be seen that the proposed resource allocation strategy supports more secondary routes than CSUM and CMIN algorithms, especially when the unoccupied channels are scarce for the usage of secondary users. Obviously, the proposed approach allocates higher priority to the secondary users with weaker link due to less idle channels. Meanwhile, the adaptable power control technique can improve the usage of cognitive idle channels by controlling the interference for the primary users and secondary users. These advantages make more multi-hop services of secondary users access the base station or access point easier.

Fig.4 The number of connective multi-hop routes varying with the number of channels

Note that the number of connective routes becomes more and more consistent when the number of unoccupied channels is large enough. Conversely, it can be explained that the performance difference in the aspect of fairness among secondary users becomes obvious when the idle channels lack of usage for so many secondary routes. Therefore, the proposed mechanism outperforms the other spectrum assignment algorithms in terms of fairness, especially in the overloaded conditions.

4 Conclusion

In this paper, have analyzed the spectrum assignment algorithms based on color-sensitive graph coloring (CSGC) model in OFDM-based multi-hop cognitive radio networks. In order to improve the spectrum efficiency and adapt the dynamic topology to a certain extent, a novel bandwidth allocation policy is proposed combined with power control strategy with the optimization objective function of maximizing the throughput of multi-hop cognitive radio network for all secondary users. Numerical results show that the proposed channel and power allocation strategy performs better than CSUM and CMIN algorithms in performances of both total network bandwidth and minimum route bandwidth of all routes, as well as the number of connected multi-hop routes. It also benefits from the consideration of fairness among secondary users, and more connected secondary routes improve the total throughput of cognitive radio networks.

Note that our proposed OFDM-based multi-hop cognitive radio networks architecture and resource allocation strategy are based on a centralized method. The future work will consider the bandwidth allocation and power control strategy in a distributed multi-hop network deployment and highly dynamic topology environment.

[ 1] Akyildiz I F, Altunbasak Y, Fekri F, et al. AdaptNet: adaptive protocol suite for next generation wireless internet. IEEE Communications Magazine, 2004, 42(3): 128-138

[ 2] Peng C, Zheng H, Zhao B Y. Utilization and fairness in spectrum assignment for opportunistic spectrum access. ACM/Springer Mobile Networks and Applications, 2006, 11(4): 555-576

[ 3] Xing Y P, Mathur C N, Chandramouli R, et al. Dynamic spectrum access with QoS and interference temperature constraints. IEEE Transactions on Mobile Computing, 2007, 6(4): 423-433

[ 4] Le L, Hossain E. QoS-aware spectrum sharing in cognitive wireless networks. In: Proceedings of the 2007 IEEE Global Communications Conference, Washington, DC, USA, 2007. 26-30

[ 5] Hoang A T, Liang Y C. Downlink channel assignment and power control for cognitive radio networks. IEEE Transactions on Wireless Communications, 2008, 7(8): 3106-3117

[ 6] Hoang A T, LianY Cg, Islam M H. Power control and channel allocation in cognitive radio networks with primary users' cooperation. IEEE Transactions on Mobile Computing, 2010, 9(3): 348-360

[ 7] Behzad A, Rubin I. Multiple access protocol for power-controlled wireless access nets. IEEE Transactions on Mobile Computing, 2004, 3(4):307-316

[ 8] Kulkarni G, Adlakha S, Srivastava M. Subcarrier allocation and bit loading algorithms for OFDMA-based wireless networks. IEEE Transactions Mobile Computing, 2005, 4(6): 652-662

[ 9] Tang J, Xue G L, Chandler C, et al. Link scheduling with power control for throughput enhancement in multihop wireless networks. IEEE Transactions on Vehicular Technology, 2006, 55(3): 733-742

[10] Shi Y, Hou Y T. Optimal power control for multi-hop software defined radio networks. In: Proceedings of the 2007 IEEE International Conference on Computer Communications, Anchorage, USA, 2007. 1694-1702

Pei Xuebing, bore in 1980. He received his Ph.D. and M.S. degrees in Department of Electronics and Information Engineering from Huazhong University of Science and Technology in 2009 and in 2006, respectively. He also received his B.S. degree in Department of Mathematics from Zhejiang University in 2003. He is currently an Engineer with China Ship Development and Design Center in Wuhan. His research interests include heterogeneous wireless networks, cognitive radio networks, and electronmagnetic compatibility in shipboard system of electronics and information.

10.3772/j.issn.1006-6748.2015.03.011

①Supported by the National Natural Science Foundation of China (No. 61461006) and the Guangxi Province Natural Science Foundation (No. 2013GXNSFBA19271).

②To whom correspondence should be addressed. E-mail: peixbhust@163.com Received on Dec. 22， 2014*, Tang Zhenhua**