Effects of compression and window size on remote acoustic identification using sensor networks

Recently the cost-benefits of automated sensing over traditional field surveys for population management of fauna has been recognised. Remote monitoring through automatic identification based on sensor networks has followed one of two approaches; using the sensor nodes to perform data analysis within the network or alternatively using the sensor network as a means for collecting data to be centrally processed. In either case a key goal is minimising power consumption in sensor nodes which imposes constraints on both processing and communication capabilities. While the first approach aims to minimise communication requirements the other aims to reduce processing requirements. In the context of sensor networks for remote monitoring utilising centralised processing, this paper considers the impact on two different strategies for reducing communication requirements on the overall system performance.