Multi-Task Wireless Sensor Network for Joint Distributed Node-Specific Signal Enhancement, LCMV Beamforming and DOA Estimation

Abstract

We consider a multi-task Wireless Sensor Network (WSN) where some of the nodes aim at applying a multi-channel Wiener filter to denoise their local sensor signals, while others aim at implementing a linearly constrained minimum variance beamformer to extract node-specific desired signals and cancel interfering signals, and again others aim at estimating the nodespecific direction-of-arrival of a set of desired sources. For this multi-task WSN, by relying on distributed signal estimation techniques that incorporate a low-rank approximation of the desired signals correlation matrix, we design a distributed algorithm under which the nodes cooperate with reduced communication resources even though they are solving different signal processing tasks and do not know the tasks of the other nodes. Convergence and optimality results show that the proposed algorithm lets all the nodes achieve the network-wide centralized solution of their node-specific estimation problem. Finally, the algorithm is applied in a wireless acoustic sensor network scenario with multiple speech sources to show the effectiveness of the algorithm and support the theoretical results. I. INTRODUCTION A Wireless Sensor Network (WSN) consists of a collection of sensor nodes with sensing, processing, and wireless communication facilities. Often, these nodes are connected in an ad-hoc fashion and then cooperate to solve a Signal Processing (SP) task by means of a distributed SP algorithm. Such distributed algorithms avoid an energy-inefficient data centralization and allow to distribute the processing load over the different nodes. Traditionally, the design of distributed algorithms has focused on WSNs where all the nodes contribute

Distributed Rate-Constrained LCMV Beamforming

“…In a centralized setting, the RTF can be estimated using covariance substraction or covariance whitening method [20]. In the distributed setting this can be estimated using [21]- [24]. Further, we assume that all sources are mutually uncorrelated, and the early reflections and late reverberation are also mutually uncorrelated (which is strictly speaking true under the assumption that the STFT coefficients S across time are uncorrelated), such that the second-order statistics (SOS) of the noise components can be written as…”

Edge Computing-Enabled Wireless Sensor Networks for Multiple Data Collection Tasks in Smart Agriculture

“…Therefore, more and more studies focus on the topic for different perspective, especially network attacks, data collection. Concerning multiple sensor data tasks, the researchers have proposed different methods to meet the requirements from hardware [12][13][14] and task scheduling and optimization [15,16]. On the sensing level, these methods can be summarized into two large groups: Periodically Sensing with All Nodes (PSAN) and Effective Node Sensing (ENS).…”

Multiple Source Location Estimation on a Dataset of Real Recordings in a Wireless Acoustic Sensor Network

“…Compared to traditional microphone arrays, WASNs can provide improved sampling of the acoustic environment, since the multiple acoustic nodes that are distributed over the monitored area increase the probability of finding a microphone that is close to the source of interest [1]. Their potential has been explored with promising results in numerous applications, such as DOA estimation [2], estimation of the exact location of the acoustic sources [3]- [7], wildlife monitoring [8], and speech enhancement and beamforming [9]- [11].…”