DOA-Based Multi-Node Geometry Calibration for Wireless Acoustic Sensor Network

In this paper, we propose a novel swarm-based method for direction of arrival (DOA) based geometry calibration of wireless acoustic sensor network (WASN). First, we formulate the objective function for multi-node geometry calibration problem by using the DOA estimations from the array nodes. Then we solve the optimization problem with randomly selected candidate solutions by whale optimization algorithm (WOA) and opposition-based learning. Experimental results show that the proposed method achieves the lowest distance error and rotation error compared with baseline methods. Further analysis demonstrates that the errors in DOA estimations caused by noise and reverberation are mitigated by the randomly selected candidate approach of WOA and opposition-based learning.