Minimizing backlog for downlink of energy harvesting networks

A transmitter powered by a renewable energy source becomes self sustainable. In this paper, we consider the broadcast channel with a transmitter and N receivers. The transmitter is powered by a renewable energy source and has finite battery capacity. Transmitter requires power P_i to transmit a packet to i^th user. In this setting, our objective is to minimize the expected backlog at the transmitter while accounting for randomness in the arrival and the recharge processes. We formulate the problem as an infinite horizon Markov Decision Process (MDP) problem and obtain the structural properties of an optimal policy. These structural properties provide valuable insights for designing close to optimal policies that are computationally efficient for real life implementations. In special cases, we provide complete description of an optimal policy.