Multi-sensor Event Detection under Temporal Correlations with Renewable Energy Sources

Sensor networks have major applications in environmental monitoring, relief operations, surveillance, health-care and defense. Future sensor networks would comprise of sensing devices with energy harvesting capabilities from renewable energy sources such as solar power. Multiple sensor nodes deployed in the region of interest would collaborate to achieve a global objective, such as detection of application specific events. This paper focuses on the design of efficient algorithms for multi-sensor activation in order to optimize the overall event detection probability. The recharge-discharge dynamics of the individual rechargeable sensor nodes, along with temporally correlated nature of event occurrences makes the optimal multi-sensor event detection question very challenging. We formulate the dynamic multi-sensor event detection question in a stochastic optimization framework, and design efficient sensor activation algorithms. Particularly, we analyze certain classes of threshold activation policies and show that they achieve near-optimal performance when the threshold is chosen carefully. Specifically, we show that a time-invariant threshold policy, which attempts to maintain a fixed number (appropriately chosen) of sensors active at all times, is optimal in absence of temporal correlations. Moreover, the same energy-balancing time-invariant threshold policy approaches optimality in presence of temporal correlations as well, albeit under certain limiting assumptions. Through simulation studies, we compare the performance of this time-invariant policy with energy-balancing correlation-dependent policies, and observe that although the latter perform better, the performance difference is rather small.