Performance Evaluation of a Cognitive Radio Network with Exponential and Truncated Usage Models

In this paper, we study the forced termination probability, blocking probability and throughput of secondary users with practical usage models, which include fixed, truncated exponential, truncated lognormal, truncated Pareto distributions. Moreover, we compare the results of these usage models with the well known exponential distribution model. We consider two system scenarios with/without spectrum handoff. For all distributions, introduction of spectrum handoff provides significant gains in forced termination probability and throughput. We found that the behaviour of all the truncated distributions is similar to exponential distribution. The results show that among all the distributions models, the secondary user connections with fixed length has the largest aggregate throughput, whereas the truncated Pareto has the smallest aggregate throughput. In general, as the mean duration of secondary user connections decreases, the throughput increases.