Cross-Layer Optimized Call Admission Control in Cognitive Radio Networks

In Cognitive Radio (CR) networks, Call Admission Control (CAC) is a key enabling technique to ensure Quality-of-Service (QoS) provisioning for Secondary Users (SUs). CAC decisions are usually made based on the current traffic volume in the system. However, in CR networks, the system state of channel utilization can only be partially observed through spectrum sensing. The presence of sensing error may mislead the CAC strategy to make an inefficient or even incorrect decision. To achieve QoS provisioning in CR networks, a practical CAC strategy should have in-built functionality to deal with the inaccuracy of sensing results. This paper is motivated to construct a cross-layer optimization framework, in which the parameters of CAC strategy and spectrum sensing scheme are simultaneously tuned to minimize the dropping rate while satisfying the requirements of both blocking rate and interference threshold. After introducing a multiple-stair Markov model to approximate the non-memoryless state transitions, the cross-layer optimization is modelled as a non-linear programming problem. The method of branch-and-bound is employed to solve the problem, where five components are involved: problem selection, reformulation linear technique, simplex method, local search and sub-problem generation. Extensive simulations are carried out to evaluate the proposed CAC strategy. The simulation results show that our CAC strategy significantly outperforms two traditional strategies. The dropping rate in our strategy is considerably reduced. Meanwhile, the blocking rate and the interference probability strictly coincide with the constraints.