An efficient policy for D2D communications and energy harvesting in cognitive radios: Go Bayesian!

Recently, there has been a surge of interests in paradigms such as device-to-device (D2D) communications and radio frequency energy harvesting (RFEH) to improve the spectrum as well as energy efficiencies of next-generation decentralized cognitive radio networks. However, little attention has been paid to the dual but competing task of subband selection of any desired bandwidth in D2D mode (i.e., opportunistic vacant spectrum access) and RFEH mode as well as need to minimize the subband switching cost (SSC) for an efficient implementation. Taking these factors into account, a new D2D-RFEH policy is proposed. It consists of: 1)Bayesian approach based Tunable Thompson Sampling (TTS) algorithm to learn subband statistics, 2) Subband access scheme employing TTS algorithm for minimizing collisions among the secondary users, and 3) Mode selection scheme. The simulation results, complexity and SSC analysis validate the superiority of the proposed policy over the policies employing frequentist approach based learning algorithms.