Secondary access policies with imperfect sensing in dynamic spectrum access networks

We consider secondary access policies with imperfect sensing in sensing-based dynamic spectrum access networks. Interference caused by both missed detections as well as the primary user´s return to the channel during the secondary transmission are explored. We allow arbitrary distributions for the primary ON/OFF intervals and expand each interval into multiple sub-intervals to create a time-varying Markov model. From this perspective, efficient and well-developed trellis-based algorithms can be applied to obtain the secondary access policies. Simulation results demonstrate that the proposed policies achieve significant improvements in the system performance tradeoff relative to existing access policies that assume the sensing is perfect.