Enabling data-centric energy-fidelity scalability in wireless body area sensor networks

Abstract

Energy-fidelity tradeoffs are central to many battery-constrained systems, but they are essential in body area sensor networks (BASNs) due to energy and resource constraints, and the critical nature of many healthcare applications. On-node signal processing and compression techniques can save energy by greatly reducing the amount of data transmitted over the wireless channel, but lossy techniques can incur a reduction in application fidelity. In order to maximize system performance, these tradeoffs must be considered at run-time due to the variable nature of BASN application, including sensed data, operating environments, user actuation, etc. BASNs therefore require energy-fidelity scalability, so automated and user-initiated tradeoff decisions can be made dynamically. This paper explores the utility of energy-fidelity scalability in BASNs from a data-centric perspective. Compression algorithms are identified that can be implemented on resource constrained BASN nodes and that have "knobs" capable of trading off compression ratios (and resulting transmission energy) with fidelity. To demonstrate the potential of energy-fidelity scalability on a real BASN and for a real application, the tradeoff space is established by adjusting these algorithms for different movement disorder data sets collected by a custom accelerometer-based BASN. Finally, mechanisms for energy-fidelity dynamic control are explored.