Joint optimization of transmission and reception policies for energy harvesting nodes

In the recent years, resource allocation problems for energy harvesting nodes have been studied extensively. While a large portion of these contributions focus on the transmit side, less attention has been paid to the receive side or jointly on both sides. In this work, we investigate the joint optimization of transmit and receive policies for a pair of energy harvesting nodes, in the basic setting of point-to-point communication over a single link. The discrete energy arrivals at the two nodes are assumed as independent Poisson processes, and are only known to the respective nodes causally. To this end, we model the system as two decentralized Markov decision processes, which do not share information about their local states but are coupled only through a global reward function, namely, the average throughput of the system. We first compute an upper bound on the system performance by assuming a central controller which is aware of the states of both nodes, and applying the policy-iteration algorithm. Then, based on the transition-independent property of the problem, we employ a low-complexity bilinear programming approach, which, via comparison with the obtained upper bound, is shown to produce local policies with very good performance.