Abstract:
We propose a modified diffusion strategy for parameter estimation in sensor networks where nodes exchange information over fading wireless channels. We show that the effe...Show MoreMetadata
Abstract:
We propose a modified diffusion strategy for parameter estimation in sensor networks where nodes exchange information over fading wireless channels. We show that the effect of fading can be mitigated by incorporating local equalization coefficients into the diffusion process. We explain how the equalization coefficients are chosen and show that the (mean) stability of the network continues to be insensitive to the choice of the combination weights and to the network topology. Our computer experiments demonstrate that the performance of the modified diffusion algorithm in fading scenario is nearly identical to that of centralized least-mean square (LMS) with equalized input data.
Date of Conference: 09-13 June 2013
Date Added to IEEE Xplore: 07 November 2013
Electronic ISBN:978-1-4673-3122-7