Real-time ML based algorithm for localizing acoustic source in WSN

• Published in: Iranian Conference on Electrical Engineering pp. 62 - 66
• Main Authors: Faraji, Mohammad Mahdi, Rezaie, Amir Hossein, Iranmehr, Ensieh
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.05.2014
• Subjects: Acoustic source localization; Acoustics; ARM processor; Collaborative algorithms; Equations; Generalized Cross Correlation; Mathematical model; Microphones; Probability function; Sensors; Signal processing algorithms; Time Delay Estimation; Wireless Sensor Network; Wireless sensor networks
• ISSN: 2164-7054
• DOI: 10.1109/IranianCEE.2014.6999504

Distributed wireless sensor networks, which are consisting of several single sensors, are becoming very popular in many important applications. This paper talks about collaborative algorithms which are implemented on WSN in order to find the location of sound source. In this paper, an algorithm is discussed for each node and a collaborative algorithm is proposed for sink node. The algorithms are designed in order to find the location of acoustic source in realtime. For localizing, 4 nodes are used. Each of them finds the direction of acoustic source individually. Then, they transfer their computed angle to sink node in order to localize acoustic source. Thus, 4 direction finder nodes and 1 sink node are designed. Direction finder nodes, which consist of 2 microphones, compute the angle of acoustic source respect to itself using GCC algorithm. Sink node receives the computed angles through wireless communication. Moreover, sink node computes the probability graph of the environment. Then, the maximum point of this graph determines the location of acoustic source. At last, represented nodes are built and they are tested in practical environment. The experimental results demonstrate that the proposed algorithm is accurate and feasible.