Seasonally aware routing for thermoelectric energy harvesting wireless sensor networks

Energy-aware routing schemes in wireless sensor networks (WSNs) often employ artificial energy assumptions, e.g., equal initial energy reserves for all nodes. Instead, we consider the case of realistic energy reserves collected via thermoelectric energy harvesting in an apartment complex and examine how the harvested energy impacts routing decisions over relatively large time frames. We formulate the corresponding multi-commodity routing flow problem and, using real observed data, remark that maximizing the volume of collected data typically leads to an uneven collection from each sensor. We propose a corresponding adjustment to the optimization problem to derive a “fair” data collection strategy. We additionally present a low overhead method of constructing a seasonally-aware routing scheme and study its performance. We compare the seasonally-aware routing performance against that of an ideal, centralized, optimization solution, as well as against a simple strategy to avoid extreme variance of residual energy at the sensor nodes.