Approximate Analysis of Cognitive Radio Systems Using Time-Scale Separation and its Accuracy

We model a cognitive radio system as a quasi-birth-death (QBD) process and determine its performance parameters. We also model the system at the quasi-stationary limiting regime. We show that this regime defines the asymptotic system behavior. The performance parameters of interest at this regime are independent of the service time distributions and can be determined by simple recursions. We propose and evaluate a new methodology to determine when the quasi-stationary approximation can be considered a good approximation of the actual system behavior. It requires low computational cost and does not require to solve the exact system.