A high-throughput wireless-powered relay network with joint time and power allocations

Abstract

This paper studies how to maximise the throughput for a wireless-powered relay network where an unreliable direct link exists between a source node and a destination node. Our idea is to optimally combine time allocations (for different relay activities) and power allocations (for the relay to forward different frames) based on a harvest-store-consume (HSC) model. With the HSC model, the network may operate in a direct transmission (DT) mode or a cooperative transmission (CT) mode to transmit one frame. Accordingly, the relay may either harvest energy from the radio-frequency signals in the direct link in DT mode, or harvest energy and forward information in CT mode, where the energy harvested in both modes is stored temporarily before it is consumed for information relaying. Moreover, in our CT mode, the relay distributes the energy to the transmissions of multiple consecutive frames by taking time-varying wireless channel conditions in a time period into account in order to achieve the best throughput performance. We formulate a deterministic optimization problem and a stochastic programming problem under the assumption of full channel state information (CSI) and causal CSI, respectively. Both problems optimally determine DT or CT modes for each frame transmission. In the CT mode, the problems also decide the time-switching (TS) ratios and the transmit power at the relay. Our formulated problems are intractable due to the energy distribution at the relay between multiple frames. This intractability is addressed by dynamic programming techniques which decompose the problems into two tractable subproblems: an outer problem, and an inner problem. By solving the two subproblems, we propose the joint TS operation and power allocation algorithms for the two original formulated problems. The proposed algorithm for causal CSI is an online algorithm with low complexity, while the proposed algorithm for full CSI is offline and provides a benchmark for the online algorithm.