Mathematical modelling of dynamic spectrum management in cognitive network

Abstract

A survey made by a Spectrum Policy Task Force (SPTF) within Federal Communications Commission (FCC) indicates that the actual licensed spectrum is largely under-utilized. A remedy to spectrum underutilization is to allow secondary users to access underutilized licensed bands dynamically when licensed users are absent. Due to this spectrum usage is undergoing a paradigm shift from the traditional licensed allocation to the dynamic spectrum access (DSA). Cognitive radios are intelligent radio, which can implement DSA efficiently. A cognitive radio can detect vacant licensed spectrum, access it and vacate when the licensed user starts transmitting. The cognitive radios can detect the spectrum more efficiently if they cooperate with other cognitive radios in the network. This results in increased transmission opportunity and hence increases the throughput of the cognitive radio network. Therefore, in order to realize the full potential of cognitive radio, there is a need for well-designed distributed cooperative algorithms that can realise the numerous gains from the vacant licensed spectrum. This thesis uses matching theory to develop such cooperation mechanism. Matching theory is a mathematical framework used to describe the formation of mutually beneficial relationships over time. This mutually beneficial relationship encourages the cognitive radio to form groups known as coalitions. The goal of matching theory in this thesis is to form coalitions of cognitive radios so that the overall benefits termed as 'utility' is improved compared to benefit that cognitive radio receives when acting alone. This improved utility due to coalition formation results in improved spectrum detection, which in turn increases the opportunities for transmission in the vacant licensed spectrum. The detected vacant spectrum is shared among the cognitive radios in the network for achieving a higher throughput. For the purpose of coalition formation using matching theory, two algorithms are proposed. The first algorithm for coalition formation uses well-known Gale-Shapely algorithm to achieve cooperation among the cognitive radios for spectrum detection and management. This algorithm results in the formation of stable coalitions of cognitive radio. In order to form coalitions, each cognitive radio prepares a preference list of other radio in the vicinity with which the cognitive radio wants to cooperate and hence form a coalition. Each cognitive radio makes an offer to cognitive radio in its preference list for cooperation. The cognitive radio can accept or reject the offer based on the preference list. The second algorithm is based one-sided matching theory which is a variant of the Gale-Shapely algorithm. The procedure is similar to the first algorithm, however, the difference is in how the coalition formation takes place among the cognitive radios. Finally, using simulations and mathematical results, various aspects of the proposed algorithms were investigate and analysed. The proposed algorithms resulted in improved spectrum detection as well as spectrum management hence enhancing the throughput of the cognitive radio network as well as increasing the spectrum efficiency. Compared to the non-cooperative scenario the modified Gale-Shapley algorithm resulted in the reduction of false alarm probability approximately by 51% and one-sided matching resulted in a 46% reduction in AWGN channel when number of cognitive radio user in the network is set to 30. While in the fading environment the reduction was approximately by 40% and 39% respectively for modified Gale-Shapley and one-sided matching algorithm.