Simulation of the performance of Cognitive Radio Networks with unreliable servers

This paper deals with a Cognitive Radio Network (CRN) which is modeled using a retrial queuing system with two finite-sources. This network includes two non-independent service units treating two types of users: Primary Users (PU) and Secondary Users (SU). The primary unit has priority queue (FIFO) and a second service unit contains an orbit both units are dedicated for the Primary Users and Secondary Users, respectively. The current work highlights the unreliability of the servers as we are assuming that both servers of this network are subject to random breakdowns and repairs. All the inter-event times in this CRN are either exponentially or non-exponentially distributed. The novelty of our investigation is to analyze the effect of several distributions (Gamma, Pareto, Log-normal, HypoExponential and Hyper-Exponential) of the failure and repair times on the main performance measure of the system. By the help of simulation we show some interesting results concerning to sensitivity problems.