Competitive Design of Energy Harvesting Communications in Wireless Fading Channels

This paper considers the design of online transmission strategies for slotted energy harvesting point-to-point communication systems in wireless fading channels. Online transmission strategies decide the amount of energy allocated to each transmission slot based on the energy harvested amounts and channel gains observed in the current and previous time slots. Offline strategies, in contrast, assume non-causal knowledge of future energy arrivals and channel gains. We adopt a worst case design objective. For a given online policy, we are interested in computing its maximum rate gap that is defined as the difference between the offline and online rates, maximized over all possible energy arrivals and channel states. The competitive rate gap is then defined as the minimum maximum rate gap over all possible online strategies. Here, we obtain, within a constant, the maximum rate gap for the Myopic policy, which equally distributes the available energy over the remaining slots, and provide an upper and a lower bound on the competitive rate gap. Moreover, we propose a new online policy targeting the competitive rate gap. Numerical results show that the policy proposed performs close to the competitive rate gap lower bound in constant and arbitrarily varying channels, and obtains good performance with real energy harvesting traces.