Compressive sensing in wireless sensor network for poultry acoustic monitoring

Abstract: A wireless acoustic sensor network was realized using wireless sensor nodes equipped with microphone condensers, in which its sensor nodes were configured to capture poultry sound data and transmit it via the network to a collection point. A high performance computer can process these large volumes of animal audio signals under different behaviors. By performing data signal processing and analyzing the audio signal, poultry sound can be achieved and then transformed into their corresponding behavioral modes for welfare assessment. In this study, compressive sensing algorithm was developed in consideration of the balance between the power saving from compression ratio and the computational cost, and a low power consumption as well as an inexpensive sensor node was designed as the elementary unit of poultry acoustic data collecting and transmission. Then, a Zigbee-based wireless acoustic sensor network was developed to meet the challenges of short transmission range and limited resources of storage and energy. Experimental results demonstrate that the compressive sensing algorithm can improve the communication performances of the wireless acoustic sensor network with high reliability, low packet loss rate and low energy consumption.