Sensing Strategies for Opportunistic Spectrum Access in Cognitive Radio Networks

Abstract

In this thesis, we address an important issue in opportunistic spectrum sensing, dealing with medium access strategies in decentralized cognitive radio networks. Speci cally, our focus is on MAC layer channel selection schemes for e cient discovery and allocation of spectrum opportunities. Opportunistic Spectrum Access (OSA) is developed as a dynamic resource allocation model to e ciently utilize the scarce resource of wireless spectrum. Particularly, low-priority Secondary Users (SUs) are allowed to share the spectrum with licensed Primary Users (PUs) in an opportunistic non-intrusive manner, such that no interference will be introduced to the PUs. This involves spectrum sensing, where SUs monitor the activity of PUs to identify and further utilize the idle bands, whenever no primary activity is detected. Recognizing hardware restrictions and the overhead caused by central infrastructure, we assume SUs have no prior knowledge about primary activity and channel state information. In this uncertain environment, secondary nodes that are cognitive devices have to distributively learn the primary activity parameters at the same time as sensing the spectrum for accessing idle bands. The goal is to maximize secondary network spectral utilization while minimizing interference introduced to the primary. This is where exploration versus exploitation dilemma arises in search for a balance between choosing empirically best channel while investigating other channels for potential opportunities. Moreover, competition should also be dealt with in order to prevent collision when multiple secondary users in the network intend to access the same channel. In this thesis, after introducing the concept of OSA for dynamic resource allocation, and discussing relevant existing work in the literature, we consider the problem of spectrum sensing and arising issues in a fading environment. Collaborative spectrum sensing is then addressed as a method to combat undesired fading e ects. Afterwards, MAC layer sensing and channel selection problem is considered. First, by modeling the problem as a multi- armed bandit, a sub-optimal channel selection algorithm referred to as modi ed-myopic strategy is proposed for the single-user scenario. Providing analysis and simulation results, we will show e cient as well as timely performance of our method compared to other strategies in the literature. Next, taking advantage of generalized Carrier Sense Multiple Access-Collision Avoidance (CSMA-CA) technique, we extend our algorithm to design a fair and low-complexity asymptotically optimal access strategy in the multi-user scenario. Analyses and simulation results are provided to evaluate the performance in dense as well as sparse networks. As a result, maximal network utilization, fairly distributed among users, is achievable in high-density decentralized network.