Sparse event detection in wireless sensor networks using analog fountain codes

In this paper, we focus on the sparse event detection (SED) problem in wireless sensor networks (WSNs), where a set of sensor nodes are placed in the field of interest to capture sparse active event sources. The SED problem in WSNs is represented from a coding theory perspective by using capacity approaching analog fountain codes (AFCs) and further solved with a standard belief propagation (BP) decoding algorithm. We show that the sensing process in WSNs produces an equivalent analog fountain code at the sink node, with code parameters determined based on the sensing capability of sensor nodes and channel gains. We analyze the probability of false detection of the proposed approach and show it to be negligible. Simulation results show that the proposed approach, namely sparse event detection with AFC (SED-AFC), achieves a significantly higher probability of correct detection (PCD) compared to existing literature at various SNR values, with a negligible probability of false detection (PFD). Moreover, we show that the minimum sampling ratio in high SNRs for the proposed scheme reaches the lower bound which is mainly characterized by the sensing coverage of the sensors and field dimensions.