Direction of Arrival Estimation in Reverberant Rooms Using a Resource-Constrained Wireless Sensor Network

A WSN capable of estimating the direction of arrival (DOA) of an acoustic source can be utilized in building surveillance, habitat monitoring or smart environment applications. In this paper, we discuss the design and implementation of a WSN that can provide an accurate DOA estimate of a single acoustic source placed in a moderately reverberant room by using only four strictly resource-constrained Berkeley Mica2 motes. We provide experimental results on the performance of our WSN in estimating the DOA of a wideband speech signal placed at different positions in the room. Using the Steered-Response-Power Phase-Transform (SRP-PHAT) algorithm along with an additional refining stage, our WSN can successfully estimate the source DOA at an 89% average success rate when the RMS DOA error margin allowed on the estimates is set to 5deg. Our results show that DOA estimation can be achieved using wireless platforms that are not as powerful as iPAQs or laptops, which were the platforms previously employed in the literature for similar purposes.