Filtering approaches to accelerated consensus in diffusion sensor networks |
| |
Authors: | Emad Abd‐Elrady Bernard Mulgrew |
| |
Affiliation: | 1. Higher Colleges of Technology, Abu Dhabi Women's College (ADWC), 41012 Abu Dhabi, UAE;2. Institute for Digital Communication, School of Engineering, The University of Edinburgh, EH9 3JL Edinburgh, UK |
| |
Abstract: | The main objective in distributed sensor networks is to reach agreement or consensus on values acquired by the sensors. A common methodology to approach this problem is using the iterative and weighted linear combination of those values to which each sensor has access. Different methods to compute appropriate weights have been extensively studied, but the resulting iterative algorithm still requires many iterations to provide a fairly good estimate of the consensus value. In this paper, different accelerating consensus approaches based on adaptive and non‐adaptive filtering techniques are studied and applied on the problem of acoustic source localization using the adaptive projected subgradient method. A comparative simulation study shows that the non‐adaptive polynomial filters based on Newton's interpolating polynomials and semi‐definite programming can provide more accelerated consensus and better estimation accuracy than adaptive filters evaluated using constrained affine projection algorithm or stochastic gradient algorithm provided that the network topology is known beforehand. Copyright © 2013 John Wiley & Sons, Ltd. |
| |
Keywords: | adaptive filtering adaptive projected subgradient method consensus diffusion networks distributed algorithms optimization methods semi‐definite programming |
|
|