Toward optimal cooperative sensing and cooperative relay in cognitive radio networks

Cognitive radio networks (CRNs) have been proposed in recent years to solve the spectrum scarcity problem by exploiting the existence of spectrum holes. This thesis studies two problems in CRNs, the cooperative spectrum sensing scheduling problem and the joint spectrum allocation and relay selection problem in CRNs. The first problem is how to optimally schedule cooperative spectrum sensing in CRNs. In CRNs, secondary users can be coordinated to perform spectrum sensing so that primary user activities can be detected more accurately. However, in a dynamic spectrum environment, more sensing cooperation may induce every secondary user to sense more channels and thus decreasing their transmission times. In Chapter 2, we study this tradeoff by using the theory of partially observable Markov decision process (POMDP). This formulation leads to an optimal sensing scheduling policy that determines which secondary users sense which channels with what miss detection probability and false alarm probability. A myopic policy with lower complexity yet comparable performance is also proposed. Numerical and simulation results are provided to illustrate that our design can utilize the spectrum more efficiently for cognitive radio users. The second problem is the joint spectrum allocation and relay selection problem in CRNs. One of the important issues in a cellular CRN is how to efficiently allocate primary user (PU) spectrums without causing harmful interference to PUs. In Chapter 3, we present a cross-layer framework which jointly considers spectrum allocation and relay selection with the objective of maximizing the minimum traffic demand of secondary users that can be supported in a cellular CRN. Specifically, we consider (1) CRN tries to utilize PU spectrums even when the CRN cell is not outside the protection region of the PU cell, and (2) cooperative relay is used in cellular CRNs to improve the utilization of PU spectrums. We formulate this cross-layer design problem as a mixed integer linear programming (MILP) and use LINGO to obtain the optimal solution. Compared with a conventional channel allocation scheme, the numerical results show a significant improvement by using our scheme