A Maximin Optimal Online Power Control Policy for Energy Harvesting Communications

A new online power control policy is constructed for a discrete-time energy harvesting communication system equipped with a battery of finite capacity. This policy only requires the knowledge of the (effective) mean of the energy arrival process and is optimal in the sense that it maximizes the minimum asymptotic expected average reward for i.i.d. energy arrival processes with the same (effective) mean. We show that the fixed fraction policy by Shaviv and O¨ zgu¨ r can be interpreted as a linear approximation of this new policy in the large batterycapacity limit and propose a locally fixed fraction policy as its more accurate approximation. A systematic comparison of these policies is conducted via analytical and numerical performance analysis.