1 Link Scheduling and Channel Assignment in Multi-channel Cognitive Radio Networks: Spectrum Underlay Approach Mui Van Nguyen and Choong Seon Hong Dept. of Computer Engineering, Kyung Hee University, 449-701, Republic of Korea Email: {nvmui, cshong}@khu.ac.kr Abstract—In this paper, we investigate the performance of multi-channel cognitive radio networks (CRNs) by taking into consideration the problem of channel assignment and link scheduling. We assume that secondary nodes are equipped with multiple radios and can switch among multiple channels. How to allocate channels to links and how much power used on each channel to avoid mutual interference among secondary links are the key problem for such CRNs. We formulate the problem of channel assignment and link scheduling as a combinatorial optimization problem. Then, we propose a the optimal solution and show that it converges to maximum optimum in some iterations by using numerical results. I. I NTRODUCTION Recently, cognitive radio [1] has been realized to be novel communication paradigm which can relax spectrum scarcity. Each node in CRNs is considered the smart wireless node which can switch and work on a different set of available frequency bands without being limited by the number of radio interfaces. The major issue of CRNs is how to effectively assign channels to links and allocate power per channel to improve spectral efficiency and obtain high overall throughput. Moreover, secondary links must align its interference to PU- Rx at acceptable target. In CRNs, the spectrum opportunity of each link is expressed as the link capacity which they can achieve on channel assigned minus the cost they must pay to use that channel. To get high spectrum opportunity, links need performing channel assignment strategies and power control policy. More impor- tantly, secondary links must cooperate together to achieve the totally perfect scheduling. In the literature, there are some studies of resource allocation for such CRNs. Shi [2] and [3] formulated the optimization problem of power and routing using the bandwidth-footprint-product (BFP) as an objective metric, aiming to minimize the interference footprint area on each channel. Thereby, the secondary nodes transmitting on the same bandwidth with suitable power levels which may not make interference to each other. A channel is allowed to use only if the secondary node senses that band idle and is not in the other node’s interference range. Consequently, their This research was supported by the MKE(The Ministry of Knowledge Economy), Korea, under the ITRC(Information Technology Research Center) support program supervised by the NIPA(National IT Industry Promotion Agency)” (NIPA-2012-(H0301-12-1004). Dr. CS Hong is the corresponding author. proposed algorithms must keep track of the set of nodes fall in the transmission range and the set of nodes that can produce interference whenever the transmit power is changed at each node. Such solutions make the implementation of the local search algorithm [2] and the distributed optimization algorithm [3] become more complicated and unscalable. In underlay fashion of spectrum sharing, secondary users’s transmssion can make harmful interference to PU’s reception. Most existing works [3], [4], and [5] applied Listen-Before- Talk (LBT) technique to detect the presence or absence of primary signals before channel access in order to avoid inter- fering with primary users (PU). However, the authors does not take the aggregate interference from multiple potential SUs’s transmission into account at PU receivers. There will be no transmission from SUs while a PU system operating under full load can tolerate more interference. To address this shortcom- ings, we propose a optimization framework for link scheduling and spectrum assignment which mutual interference among secondary users are relaxed by protocol interference model while aggregate interference caused by them to primary links is limited below a acceptable threhold. Our objective is to allocate the less power to links close to PUs. Then, we seek a combination of link-band which their total weight is maximum. II. SYSTEM MODEL AND PROBLEM FORMULATION SU3-Rx 1 P x 3 PU1-Rx PU2-Tx PU1-Tx PU2-Rx SU1-Tx SU2-Tx SUl-Rx SUl-Tx : primary link SU1-Rx SU3-Tx f_2 f_2 f_1 SU2-Rx f_m f_1 f_1 3 P 4 P PU3-Rx PU3-Tx f_3 : secondary link l P Fig. 1: Co-existence of SUs and PUs in CRNs We consider MHCRNs modeled by the set of secondary links L as illustrated in Fig. 1. We assume that the whole spectrum is divided into a set of orthogonal frequency channels 2012 한국컴퓨터종합학술대회 논문집 Vol.39, No.1(D)