Abstract
The problem of opportunistically accessing a set of parallel channels occupied by primary users is considered. A continuous-time Markov chain is introduced to model channel occupancy of the primary users. A slotted transmission protocol using a periodic sensing strategy with optimal dynamic access is proposed. To maximum channel utilization while limiting interference to the primary users, a framework of constrained Markov decision processes is developed to derive the optimal access policy. A solution based on linear programming is presented. Simulations are used for performance evaluation. It is demonstrate that periodic sensing yields negligible performance loss when the constraint on interference is strict.