Context-Aware Analysis Scheduling in Wireless Body Area Networks

摘要

Wireless body area networks have seen a rise in popularity both for clinical and personal use, but, much work needs to be done to address clinical settings that are dynamic in nature. We consider the scenario where a mobile device receives several signals from various body sensors, and executes analysis in real-time. There can be periodic analysis input statically with a user-defined priority ranking, but also user-requested analysis, and event detection arising dynamically. This can be modelled as a real-time system scheduling problem, with both periodic and aperiodic tasks. Battery management is also a crucial component in medical applications, as physicians may use a device for several hours without charging. In the proposed scheduling design, we incorporate battery adaptations interleaving periods of inactivity into our schedule to preserve battery life. In a simulated scenario, we compare an adapted baseline algorithm, earliest deadline first, with a heuristic based context-aware scheduling algorithm, to determine utility in terms of completion count, effective processor utilization, and time-based valuation.