Energy efficient spatiotemporal threshold level detection in large scale wireless sensor fields

A class of energy efficient schemes is proposed for detection and tracking of the boundaries of correlated spatial distributions, such as air pollution in large cities at a known threshold level, using large scale wireless sensor network. It is shown that without having the exact statistics of the signal, distributed collaborative space and time oriented filtering can be used to efficiently detect the boundary of the threshold level of a correlated spatial distribution over time. Collaborative signal processing is used to reduce the effect of noisy observations and distributed space and time domain filtering is applied for energy conservative threshold level tracking. It is shown that by using spatial and time oriented filtering, the operating mode of the wireless sensor network is shifted from communication dominant mode toward computation and sensing dominant mode, which significantly saves the in-network energy. The performance of the discussed approach for a few correlated random spatial distributions is evaluated using computer simulations.