Robust distributed filtering for sensor networks under parametric uncertainties

• Autor(a) principal: Rocha, Kaio Douglas Teofilo
• Data de Publicação: 2022
• Tipo de documento: Tese
• Idioma: eng
• Título da fonte: Biblioteca Digital de Teses e Dissertações da USP
• Texto Completo: https://www.teses.usp.br/teses/disponiveis/18/18153/tde-17022023-123234/
• topic: Consenso; Consensus; Distributed filtering; Filtragem distribuída; Filtro de Kalman; Kalman filter; Redes de sensores; Sensor networks; Sistemas com incertezas; Uncertain systems

In the past few years, we have witnessed the rapid popularization of networked cooperative multi-agent systems, which consistently move towards becoming ubiquitous in our society. As one of the most well-established examples of such systems, sensor networks have been applied to increasingly more complex systems, demanding even more robust, efficient, and reliable technologies. Distributed state estimation is the most fundamental task that one can accomplish with these networks. The main objective of this thesis is to develop robust distributed filtering strategies for sensor networks applied to linear discrete-time systems subject to model parametric uncertainties. Specifically, we deal with two types of uncertainties: norm-bounded and polytopic. To achieve this goal, we also address other related problems, divided into two categories. The first category of problems refers to the single-sensor state estimation task. Within this category, we consider the scenarios in which the underlying models are perfectly known and where they are subject to each of the two kinds of uncertainty. We propose nominal and robust filters for each situation. The second category concerns the networks with multiple sensors, considering the same three scenarios. For each one, we propose both centralized and distributed estimators. We use the average consensus algorithm to obtain the distributed filters, which approximate their centralized counterparts. The proposed filters are based on the celebrated Kalman filter and present a similar recursive and relatively simple structure. We evaluate the performance of the proposed estimators with application examples, in which we also compare them to existing strategies from the related literature.