Adaptive multiresolution sampling in event-driven WSNs

Abstract

In time driven applications we are often bound to make a keen compromise between the sampling resolution (or sampler architecture) and available resources. This is especially true in WSNs (Wireless Sensor NetworkS), although the complexity of event driven environments present us new opportunities for optimizations. In this paper we propose a multi resolution sampling protocol for event driven WSNs that can dynamically adapt to event signatures and can efficiently schedule resource intensive samplers. The system can monitor the environment by a resource optimized sampler which is only to detect events and not to store or process any samples. When scheduled the high resolution sampler is activated to sample the appropriate physical property of the phenomenon marked by the event, via resolution optimized hardware. Since it takes time to detect an event (triggered by the signal we already ought to sample) and to power up the high resolution, resource intensive sampler, we must start the sampling before the signal is detected. This calls for dynamic event forecasting, which is the main feature of the proposed system.