In this paper, we study a distributed parameter estimation problem in a large-scale network of communication sensors. The goal of the sensors is to find a global estimate of an unknown parameter minimizing, which minimizes some aggregate cost function. Each sensor can communicated to a few “neighbors”, furthermore, the communication channels have limited capacities. To solve the resulting optimization problem, we use a weighted modification of the distributed consensus-based SPSA algorithm whose main advantage over the alternative method is its ability to work in presence of arbitrary unknown-but-bounded noises whose statistical characteristics can be unknown. We provide a convergence analysis of the weighted SPSA-based consensus algorithm and show its efficiency via numerical simulations.

Convergence Analysis of Weighted SPSA-based Consensus Algorithm in Distributed Parameter Estimation Problem / Sergeenko, Anna; Erofeeva, Victoria; Granichin, Oleg; Granichina, Olga; Proskurnikov, Anton. - ELETTRONICO. - 54:(2021), pp. 126-131. ((Intervento presentato al convegno 19th IFAC Symposium on System Identification SYSID 2021 tenutosi a Virtuale (formalmente Padova, Italia) nel 13-16 July 2021 [10.1016/j.ifacol.2021.08.346].

Convergence Analysis of Weighted SPSA-based Consensus Algorithm in Distributed Parameter Estimation Problem

Proskurnikov, Anton
2021

Abstract

In this paper, we study a distributed parameter estimation problem in a large-scale network of communication sensors. The goal of the sensors is to find a global estimate of an unknown parameter minimizing, which minimizes some aggregate cost function. Each sensor can communicated to a few “neighbors”, furthermore, the communication channels have limited capacities. To solve the resulting optimization problem, we use a weighted modification of the distributed consensus-based SPSA algorithm whose main advantage over the alternative method is its ability to work in presence of arbitrary unknown-but-bounded noises whose statistical characteristics can be unknown. We provide a convergence analysis of the weighted SPSA-based consensus algorithm and show its efficiency via numerical simulations.
File in questo prodotto:
File Dimensione Formato  
1-s2.0-S2405896321011204-main.pdf

accesso aperto

Tipologia: 2a Post-print versione editoriale / Version of Record
Licenza: Creative commons
Dimensione 679.73 kB
Formato Adobe PDF
679.73 kB Adobe PDF Visualizza/Apri
Pubblicazioni consigliate

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11583/2924392