COORDINATED SENSING OF NETWORKED BODY SENSORS USING MARKOV DECISION PROCESSES

This article describes a Markov decision process (MDP) framework for coordinated sensing between correlated sensors in a body-area network. The technique is designed to extend the life of mobile continuous health-monitoring systems based on energy-constrained wearable sensors. The technique enables distributed sensors in a body-area network to adapt their sampling rates in response to changing criticality of the detected data and the limited energy reserve at each sensor node. The relationship between energy consumption, sampling rates, and utility of coordinated measurements is formulated as an MDP. This MDP is solved to generate a globally optimal policy that specifies the sampling rates for each sensor for all possible states of the system. This policy is computed offline before deployment and only the resulting policy is stored within each sensor node. We also present a method of executing the global policy without requiring continuous communication between the sensors. Each sensor node maintains a local estimate of the global state. Communication occurs only when an information-theoretic model of the uncertainty in the local state estimates exceeds a predefined threshold. We show results on simulated data that demonstrate the efficacy of this distributed-control framework and compare the performance of the proposed controller with other policies.