Efficient accelerometer-based event detector in wireless sensor networks

In this paper an autocovariance-based event detector algorithm is proposed. The algorithm is able to detect events even if the measurements have poor signal-to-noise ratio, and its performance is independent of the characteristic of the input signal. An efficient implementation of the algorithm is also proposed, which allows the utilization of the algorithm on low-end devices, e.g. in wireless sensor networking nodes. The performance of the algorithm has been tested and compared to a conventional filter-based approach, in a vehicle detector application.