POMDP-Based Energy Cooperative Transmission Policy for Multiple Access Model Powered by Energy Harvesting

Abstract

To efficiently use harvested energy, we often consider energy cooperation between transmitter nodes in a multiple access wireless network. However the imperfect observations of current network states will degrade the energy transfer policy. In addition, it is difficult to derive an optimal data transmission policy based on the imperfect observation states. In order to solve these problems, we design a partially observed Markov decision process based energy cooperative transmission policy to maximize long-term system throughput. This policy optimizes the energy transfer and data transmission when we cannot observe energy harvesting states and channel states in a current time interval. We adopt the piecewise linearity of the value function and incremental pruning algorithm to obtain the optimal policy for all states in three energy harvesting scenarios. Simulation results show that the proposed cooperative optimal policy converges fast and produces higher sum throughput compared with other policies. Specifically, this policy could be used in device-to-device communication and vehicular networks.