Centralized Adaptation for Parameter Estimation over Wireless Sensor Networks

TL;DR

This paper introduces a novel centralized LMS algorithm for wireless sensor networks that mitigates channel impairments and bias, improving parameter estimation accuracy through data refinement and link failure detection.

Contribution

It proposes a new CLMS algorithm with data refinement, link failure alarms, and bias elimination to enhance estimation performance over fading channels.

Findings

Improved steady-state mean-square error with bias elimination.

Effective link failure detection reduces data distortion impact.

Theoretical results validated by simulations.

Abstract

We study the performance of centralized least mean-squares (CLMS) algorithms in wireless sensor networks where nodes transmit their data over fading channels to a central processing unit (e.g., fusion center or cluster head), for parameter estimation. Wireless channel impairments, including fading and path loss, distort the transmitted data, cause link failure and degrade the performance of the adaptive solutions. To address this problem, we propose a novel CLMS algorithm that uses a refined version of the transmitted data and benefits from a link failure alarm strategy to discard severely distorted data. Furthermore, to remove the bias due to communication noise from the estimate, we introduce a bias-elimination scheme that also leads to a lower steady-state mean-square error. Our theoretical findings are supported by numerical simulation results.