Online Scheduling for Energy Harvesting Two-Way Channels with Processing Costs

Abstract

We consider an energy harvesting two-way channel with processing costs, where the transmitters incur a processing cost per unit time as long as they are on for transmitting or receiving data. Such processing costs force users to perform bursty transmissions. We develop an online power scheduling algorithm where the transmitters determine their transmit powers and burst durations based only on the causal knowledge of the energy arrivals. We consider the case where the users harvest energy from a common energy source, however, the energy intakes of the users are different due to their different battery sizes. Focusing on independent and identically distributed (i.i.d.) energy arrival processes, we determine the exactly optimum power schedule for Bernoulli energy arrivals and develop a nearly optimum power schedule for general energy arrivals. Our proposed policy is distributed, which users can apply independently with no need for cooperation or coordination between them. We show that the proposed policy is within a constant multiplicative and additive gap from the optimal online policy for all energy arrivals.