Parallel aggregated MAB framework for source selection in multi-antenna RF harvesting circuit

Recently, the circuits capable of RF energy harvesting (RFEH) simultaneously from multiple dedicated RF sources have been developed. To enable RFEH from any ambient and unknown RF sources (operating in distinct frequency bands), such circuits need the capability to characterize RF sources. In this paper, a new characterization and selection policy (CSP) using parallel and aggregated multi-armed bandit (PA-MAB) framework has been proposed to aid optimum RF source selection in multi-antenna RFEH circuits. The novel contributions of this paper are, 1) Parallel framework which allows simultaneous selection of multiple optimal arms (i.e. RF sources), 2) Bayesian approach based Thompson Sampling algorithm to accomplish the characterization and selection tasks, 3) Bayesian inference to estimate the RF potential of individual band from the aggregated energy harvested from multiple chosen bands, and 4) Tunable RFEH duration to minimize the number of frequency band switching and increase RFEH duration. Derived theoretical performance bounds, as well as simulation results, validate the superiority of the proposed CSP over existing CSPs in stationary as well as quasi-stationary spectrum environments.