Multi-bit Quantizer Design for Distributed Parameter Estimation

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

We consider sensors deployed in diverse locations measuring a common parameter through noisy observations. These observations are quantized to be sent to a fusion center doing the estimation of the common parameter. We design these quantizers to minimize the worst-case mean square error for common parameter estimation. Relying on an asymptotic regime in terms of sensors' number and on random multi-bit quantizer per sensor, we provide a relevant continuous distribution for the thresholds of these quantizers via signomial programming. Through numerical simulations, we show that the proposed quantizers outperform the uniformly-distributed one and some deterministic ones even when the number of sensors is limited.

Original languageEnglish
Title of host publication2025 IEEE Statistical Signal Processing Workshop, SSP 2025
PublisherIEEE Computer Society
ISBN (Electronic)9798331518004
DOIs
Publication statusPublished - 1 Jan 2025
Event2025 IEEE Statistical Signal Processing Workshop, SSP 2025 - Edinburgh, United Kingdom
Duration: 8 Jun 202511 Jun 2025

Publication series

NameIEEE Workshop on Statistical Signal Processing Proceedings
ISSN (Print)2373-0803
ISSN (Electronic)2693-3551

Conference

Conference2025 IEEE Statistical Signal Processing Workshop, SSP 2025
Country/TerritoryUnited Kingdom
CityEdinburgh
Period8/06/2511/06/25

Keywords

  • Cramer-Rao bound
  • Distributed estimation
  • Minimax
  • Quantization
  • Signomial programming

Fingerprint

Dive into the research topics of 'Multi-bit Quantizer Design for Distributed Parameter Estimation'. Together they form a unique fingerprint.

Cite this