Energy Harvesting Networks with Energy Cooperation: Procrastinating Policies

TL;DR

This paper investigates optimal energy transfer and power policies in energy harvesting networks with energy cooperation, showing that delaying transfers until needed maximizes throughput and proposing algorithms with water-filling interpretations.

Contribution

It introduces procrastinating energy transfer policies that are optimal for both infinite and finite battery cases, simplifying joint optimization into manageable subproblems.

Findings

Procrastinating policies are sum-throughput optimal.

Bi-directional energy cooperation improves throughput.

Algorithms have water-filling interpretations.

Abstract

This paper considers multiterminal networks with energy harvesting transmitter nodes that are also capable of wirelessly transferring energy to or receiving energy from other nodes in the network. In particular, the jointly optimal transmit power and energy transfer policies that maximize sum-throughput for the two-way, two-hop, and multiple access channels are identified. It is shown for nodes with infinite-sized batteries that delaying energy transfers until energy is needed immediately at the receiving node is sum-throughput optimal. Focusing on such procrastinating policies without loss of optimality, the stated joint optimization problem can be decomposed into energy transfer and consumed energy allocation problems which are solved in tandem. This decomposition is shown to hold for the finite-sized battery case as well, using partially procrastinating policies that avoid battery overflows. It is observed that for the two-hop channel, the proposed algorithm has a two fluid water-filling interpretation, and for the multiple access channel, it reduces to a single transmitter problem with aggregate energy arrivals. Numerical results demonstrate the throughput improvement with bi-directional energy cooperation over no cooperation and uni-directional cooperation.