Latency-Optimum NOMA With RF Energy Harvesting and Finite Battery Capacity

We consider an uplink non-orthogonal multiple access (NOMA) setup in which two nodes rely on a radio-frequency energy harvester with a finite-capacity battery for their transmissions. We investigate the effect of the harvester’s battery capacity on the average latency required for the successful delivery of messages in a Rayleigh faded environment. Under a successive interference cancellation decoding strategy with an automatic-repeat request protocol, we formulate the average latency as a function of both the battery “charging time” and the successful “decoding time”. We derive a closed-form expression for the probability of outage and identify that the battery capacity induces a trade-off between the charging and the successful decoding time. Finally, we minimize the latency by iteratively optimizing the battery capacity and the energy allocation between the two transmitting nodes.