SOURCE IEEE Trans. Vehicular Tech, vol. 64, no. 8, pp. 3675-3688, Aug. 2015
This paper considers the power allocation of a single multiple-input–multiple-output (MIMO) wireless link with hybrid energy harvesting and grid power supply. We optimize the power allocation to minimize the power grid energy consumption while guaranteeing users’ quality of service (QoS), which is defined as transmitting a certain number of bits in finite time horizon. First, we solve the offline throughput maximization problem and propose the optimal solution called two-stage spatial–temporal water filling (ST-WF), which allocates harvested energy in the first stage and then power grid energy in the second stage. Based on this, we solve the offline power grid energy minimization problem by utilizing its equivalence to the throughput maximization problem. Then, we use the Markov decision process (MDP) to get the optimal online power grid energy minimization policy. Furthermore, we propose two near-optimal algorithms with reduced complexity, namely, the constant water level algorithm and the bit-aware water level algorithm. We evaluate the performance of the proposed offline and online algorithms by numerical simulations, showing that the bit-aware water level algorithm performs closely to the optimal online solution.
Index Terms—Energy harvesting, green communications, Markov decision process (MDP), multiple-input–multiple-output (MIMO) systems, two-stage spatial–temporal water filling (ST-WF).