LAUSR

This paper proposes a continuous-time communication model of an energy harvesting device (EHD) in the scenario that such EHD suffers from the temporal death caused by the energy depletion and completion death caused by the destruction of its hardware or software, respectively. The harvested energy is modeled as continuous fluid process, it arrives continuously and varies in time. The data is assumed to be infinite backlog, and the EHD transmits it via a wireless channel fluctuates randomly due to fading. We reformulate the system model into an equivalent piece-deterministic Markov process (PDMP) based on the imbedded discrete-time decision epoch sequence of the system model, and then an infinite horizon discrete-time Markov decision process (MDP) is built. We show the existence of the stationary deterministic optimal transmission power rate (TPR) policy, and an algorithm for computing the TPR policy and the maximum total expected throughput is developed. Finally, numerical examples are provided to confirm the analytical findings. The effects of some system parameters to the optimal TPR policy and the maximum expected throughput are investigated numerically.