Distributed Blind Source Separation Based on FastICA

2024-10-24

IEEE Signal Processing Letters

AI Summary

To address the challenges of energy and bandwidth constraints in wireless sensor networks—where raw data transmission is infeasible—this paper proposes a distributed blind source separation (BSS) method that eliminates the need for global prewhitening. Built upon the distributed adaptive signal fusion (DASF) framework, the approach implicitly avoids inter-node covariance computation and integrates an enhanced FastICA algorithm with local whitening and joint non-Gaussianity maximization, enabling decentralized iterative optimization. Theoretically, it guarantees recovery of the Q most non-Gaussian independent components, where Q scales linearly with communication overhead. Experiments demonstrate that, without transmitting raw data, the method achieves separation performance comparable to centralized FastICA while reducing communication cost by over 60%. To the best of our knowledge, this is the first provably convergent distributed independent component analysis (ICA) solution operating without global prewhitening.

Technology Category

• Constraint Satisfaction and Optimization: Distributed CSP/Optimization
• Search and Optimization: Distributed Search
• Machine Learning: Semi-Supervised Learning

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Experiences and lessons learnt from Web-based algorithms and system deployments
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Security and Privacy: Data transparency and provenance

Abstract

With the emergence of wireless sensor networks (WSNs), many traditional signal processing tasks are required to be computed in a distributed fashion, without transmissions of the raw data to a centralized processing unit, due to the limited energy and bandwidth resources available to the sensors. In this paper, we propose a distributed independent component analysis (ICA) algorithm, which aims at identifying the original signal sources based on observations of their mixtures measured at various sensor nodes. One of the most commonly used ICA algorithms is known as FastICA, which requires a spatial pre-whitening operation in the first step of the algorithm. Such a pre-whitening across all nodes of a WSN is impossible in a bandwidth-constrained distributed setting as it requires to correlate each channel with each other channel in the WSN. We show that an explicit network-wide pre-whitening step can be circumvented by leveraging the properties of the so-called Distributed Adaptive Signal Fusion (DASF) framework. Despite the lack of such a network-wide pre-whitening, we can still obtain the <inline-formula><tex-math notation="LaTeX">$Q$</tex-math></inline-formula> least Gaussian independent components of the centralized ICA solution, where <inline-formula><tex-math notation="LaTeX">$Q$</tex-math></inline-formula> scales linearly with the required communication load.

Problem

• Distributed signal processing in wireless sensor networks
• Eliminating network-wide pre-whitening in FastICA
• Achieving centralized ICA solution with reduced communication

Innovation

• Distributed FastICA algorithm for WSNs
• Eliminates network-wide pre-whitening step
• Uses DASF framework for signal fusion