Adaptive Sensing Schedule for Dynamic Spectrum Sharing in Time-varying Channel

Dynamic spectrum sharing is considered as one of the key features in the next-generation communications. In this correspondence, we investigate the dynamic tradeoff between the sensing performance and the achievable throughput, in the presence of time-varying fading (TVF) channels. We first establish a unified dynamic state-space model (DSM) to characterize the involved dynamic behaviors, where the occupancy states of primary user (PU) and the fading channel gains are modeled as two Markov chains. On this basis, a promising dynamic sensing schedule framework is proposed, whereby the sensing duration is adaptively adjusted based on the estimated real-time TVF channel. We formulate the sensing-throughput tradeoff problem mathematically, and further show that there exists the optimal sensing duration maximizing the throughput for the secondary user (SU), which will change dynamically with channel gains. Relying on our designed recursive sensing paradigm that is able to blindly acquire varying channel gains as well as the PU states, the sensing duration can be then adjusted in line with the evolving channel gains. Numerical simulations are provided to validate our dynamic sensing schedule algorithm, which can significantly improve the SU's throughput by reconfiguring the sensing duration according to dynamic channel conditions.