Exploiting temporal and spatial correlation in wireless sensor networks

Motivated by applications of wireless sensor networks to seismic field monitoring, we propose a method that integrates in-situ lightweight temporal compression with random access communication and compressive sensing for recovery of spatially-sparse phenomena. This method of spatio-temporal compression offers savings in terms of energy consumption and bandwidth usage, does not require sensors to be synchronized, and requires minimal feedback from the fusion center. Furthermore, the method is robust to node failures and packet losses. Performance is quantified using both simulation and real data, showing significant improvements in energy and bandwidth efficiency over more conventional techniques.