Adaptive data compression for energy harvesting wireless sensor nodes

Energy availability is one of the main constraints in the design of wireless sensor nodes for the monitoring of water distribution systems. Harvesting energy from the ambient environment has received increasing attention in modern research due to the fact that it can significantly prolong the lifetime of sensor nodes. However, power management is still a critical issue because power generation rates are random and vary over time. Data compression is a powerful tool for use in reducing the energy consumption by the sensor by reducing the number of transmitted bits. Additionally, adaptive data compression can address the trade-offs between data quality and energy consumption. In this paper, we propose a framework for adaptive data compression based on prediction techniques to adapt to energy harvested power generation. Our objective is to minimize the average distortion of the compressed data in the long run under the energy variations. We optimize this objective by tuning the compression algorithm subject to energy availability. The problem of optimizing the desired tradeoffs between data quality and energy saved is subject to power availability and event criticality is formulated and solved via a Markov Decision Process (MDP).