Optimal Selection of False Alarm Probability for Dynamic Spectrum Access

This letter investigates the selection of false alarm (FA) probability for dynamic spectrum access (DSA). We suggest that the performance of DSA systems can be improved efficiently by properly selecting the value of FA probability. The problem of DSA process with imperfect sensing is modeled as a continuous-time Markov chain and based on which, an optimization problem is formulated to maximize the number of served secondary radios for long-run term. We also present a simple gradient-based algorithm to find the optimal FA probability. Numerical results show that the proposal has better performance than conventional constant false alarm ratio based scheme.