Decision making policy for RF energy harvesting enabled cognitive radios in decentralized wireless networks

Abstract

Cognitive radio (CR) paradigm with radio frequency energy harvesting (RFEH) have significant potential to improve the network throughput by utilizing vacant spectrum using battery-operated self-sustainable radio terminals. Research efforts relevant to these paradigms are focused on the mode selection policies which decide when to switch from CR mode (i.e., opportunistic vacant spectrum access mode) to the RFEH mode (i.e., battery charging using ambient RF energy) and vice-versa. So far, very little attention has been paid to the dual but competing task of frequency band selection in CR and RFEH modes under partially observable environment in the decentralized wireless networks. Furthermore, the need of tunable bandwidth frequency band access for CRs and lower subband switching cost (SSC) for energy efficient implementation have made the design of the decision making policy (DMP) more challenging. In this paper, a new CR-RFEH DMP has been proposed for RFEH enabled CR terminals in the decentralized wireless networks. The proposed DMP consists of three sub-units: 1) Bayesian approach based tunable Thompson sampling algorithm for subband statistics estimation, 2) Thompson sampling algorithm based subband access scheme exploiting the past collision events to minimize collisions among CRs, and 3) Mode selection scheme. Simulation results show that the proposed DMP offers 10–35% improvement in the throughput of the decentralized network and 40–90% reduction in the number of subband switchings compared to existing DMPs. The simulation results are then validated using real radio signals on the proposed USRP testbed.