CMStatistics 2017: Start Registration
View Submission - CMStatistics
Title: Distributed estimation in networked systems with random parameter matrices and transmission packet dropouts Authors:  Raquel Caballero-Aguila - Universidad de Jaen (Spain)
Aurora Hermoso-Carazo - Universidad de Granada (Spain) [presenting]
Josefa Linares-Perez - Universidad de Granada (Spain)
Zidong Wang - Brunel University London (United Kingdom)
Abstract: Over the last few decades, the estimation problem in networked systems has attracted increasing attention and different kinds of fusion estimation algorithms have been reported. Recently, one of the most challenging and fertile research fields is the distributed estimation problem in sensor networks, where the sensor nodes are spatially distributed according to a given network topology, so every sensor is only connected and exchange information with its neighbors. In this framework, the aim concerns the distributed filtering problem in discrete-time stochastic systems over a sensor network with a given topology, assuming that the measured outputs are perturbed by random parameter matrices and correlated additive noises and, moreover, random packet dropouts may occur during the data transmission among the sensor nodes through the different network communication channels. By an innovation approach, using the last measurement received if a data packet is lost, an intermediate least-squares filter is designed at each sensor node using its own measurements and those from its neighboring nodes. After that, every sensor collects the intermediate filters that are successfully received from its neighbors and the final distributed filter is obtained as the least-squares matrix-weighted linear combination of the intermediate ones within its neighborhood.