Graph-Based Robust Resource Allocation for Cognitive Radio Networks

Abstract

Cognitive radio (CR) technology is promising for next generation wireless networks. It allows unlicensed secondary users to use the licensed spectrum bands as long as they do not cause unacceptable interference to the primary users who own those bands. To efficiently allocate resources in CR networks, stable resource allocation based on graph theory is investigated, which takes all users' preferences into account. In this paper, we focus on improving robustness of the stable matching based resource allocation. A truncated scheme generating almost stable matchings is first investigated. Based on the properties of the truncated scheme, two types of edge-cutting algorithms, called direct edge-cutting (DEC) and Gale-Shapley based edge-cutting (GSEC), are developed to improve resource allocation robustness to the channel state information variation. To mitigate the problem that certain secondary users may not be able to find suitable resources after edge-cutting, multi-stage (MS) algorithms are then proposed. Numerical results show that the proposed algorithms are robust to the channel state information variation.