Energy-efficient scheme for target recognition and localization in wireless acoustic sensor networks

Abstract

The development of wireless acoustic sensor networks has driven the use of acoustic signals for target monitoring. Most monitoring applications require continuous network connectivity and data transfers, which can rapidly exhaust nodes’ energy. Consequently, sensors must collaborate in an adequate architecture to perform target recognition and localization tasks and then to send the results to a remote server with a reduced data volume. The design of an energy-efficient scheme that achieves acoustic target recognition and localization remains an open research problem. Accordingly, this article proposes a low-energy acoustic-based sensing scheme for target recognition and localization to be implemented in a cluster-based sensing approach designed to appropriately balance energy consumption and local processing performed by sensor nodes. A reduced set of low-complexity feature extraction methods in the time domain signal are used in the recognition process. The scheme uses the received energy of the acoustic signals for the target localization. This article details the network architecture, the scheme specification, and its implementation. The results show that the scheme can classify targets with 81.34% accuracy. It requires 3.2 mJ of energy when executed in MICAz, achieving 99% energy savings compared to streaming 3 s of an acoustic signal to a remote server.

Design of a New Lightweight Accurate Grid-Based Localization Method for WASN

“…(a) Single acoustic target localization algorithms: Most of the energy-based acoustic source localization research works have utilized least square (LS) methods such as weighted LS [14], unconstrained LS-quadratic-term elimination (QE) [15] [16] [10], semidefinite programming (SDP) [17], projection onto convex (POCs) [18], and a diffusionbased projection [19]. solutions are proposed using either expectation maximization, multiresolution projection [9], alternating projection search method [24], or sparse approximation [25].…”

“…More specifically, Least Square estimators are sensitive to local optima, saddle points, whereas Maximum Likelihood estimators require costly iterative search and good initial region to be satisfied; otherwise, divergence and local convergence problems would happen [5]. Additionally, the prior published studies have not examined the performance of their proposed algorithms in terms of energy consumption except in [10], where localization accuracy is severely sacrificed for obtaining a low energy-cost approach. Thus, an efficient but lightweight energy-based acoustic source localization solution still needs further research.…”

Lightweight feature extraction method for efficient acoustic-based animal recognition in wireless acoustic sensor networks

“…Due to the cost-intensive algorithms used in the domain of signal processing, only a little attention was given to acoustic-based target recognition in WASNs [3,4,[10][11][12][13][14]19]. In [14], two spectral features were used in the proposed approach for binary classification of the sound to either speech or music.…”

“…Since the background noise level might vary, the silence threshold for each sound record is calculated adaptively based on the signal's average energy. In our approach, we defined the silent frame as a frame whose Rootmean-square value (RMS i ) is below 10% of the long-term average signal energy [19]. Figure 2a-c shows the result of a sound record's silence removal algorithm in the dataset.…”

“…In our previous work [19], we studied the efficiency of on-sensor acoustic target recognition and localization using time-domain features. Despite the low-complexity of the proposed methods for feature extraction, the results have shown a considerable overlapping of the extracted features for different animal classes, which increases the difficulty of classification.…”