Bayesian methods for distributed estimation in cooperative networks

• Grant number: 18/26191-0
• Support type: Regular Research Grants
• Duration: August 01, 2019 - July 31, 2021
• Field of knowledge: Engineering - Electrical Engineering - Telecommunications
• Principal researcher: Marcelo Gomes da Silva Bruno
• Grantee: Marcelo Gomes da Silva Bruno
• Home Institution: Divisão de Engenharia Eletrônica (IEE). Instituto Tecnológico de Aeronáutica (ITA). Ministério da Defesa (Brasil). São José dos Campos , SP, Brazil
• Assoc. researchers: Claudio José Bordin Júnior ; Stiven Schwanz Dias

Abstract

We propose in this project new Bayesian algorithms for distributed estimation in cooperative networks using decentralized filters that operate without a global data fusion center. In the proposed architecture, each network node records and independently processes local measurements, but different nodes are also capable of communicating with each other via message passing to build a global cooperative estimate over the network of a hidden state vector of interest. The goal, which is consistent with the state of the art in the area, is to develop fully distributed and scalable algorithms which operate on partially connected networks and approximate the optimal centralized estimate, but, at the same time, have a low internode communication cost. To reach that goal, we investigate two diffusion methods using respectively the Adapt-and-Combine (ATC) and Random Exchange (RndEx)techniques, which are formulated in a Bayesian perspective that is more general than the traditional formulation in the literature and enables the implementation of those methods using particle filters in scenarios where arbitrary nonlinear and non-Gaussian state space models are assumed. As an extension of this work, we also intend to generalize the proposed algorithms to state vectors that are defined on nonlinear topological spaces (manifolds)such as hyperspheres, as opposed to linear Euclidean spaces. Finally, by extending the aforementioned techniques to message passing algorithms in more general space-time graphs,we propose to consider the problem where multiple mobile agents linked by a partially connected network cooperatively estimate their own position and, at the same time, also cooperatively track another non-cooperative network node. Possible applications of practical interest for our work include identification of digital communication channels using cooperative filter networks and surveillance of large buildings and critical infrastructure using multiple intelligent unmanned aerial vehicles (UAVs). (AU)

Scientific publications

(References retrieved automatically from Web of Science and SciELO through information on FAPESP grants and their corresponding numbers as mentioned in the publications by the authors)

OLIVEIRA, HALLYSSON; DIAS, STIVEN SCHWANZ; BRUNO, MARCELO GOMES DA SILVA. Cooperative Terrain Navigation Using Hybrid GMM/SMC Message Passing on Factor Graphs. IEEE TRANSACTIONS ON AEROSPACE AND ELECTRONIC SYSTEMS, v. 56, n. 5, p. 3958-3970, OCT 2020. (18/26191-0)

FERNANDES, GUILHERME C. G.; DIAS, STIVEN S.; MAXIMO, MARCOS R. O. A.; BRUNO, MARCELO G. S.. Cooperative Localization for Multiple Soccer Agents Using Factor Graphs and Sequential Monte Carlo. IEEE ACCESS, v. 8, p. 213168-213184, 2020. (18/26191-0)

DE FIGUEREDO, CAIO G.; BORDIN JR, CLAUDIO J.; BRUNO, MARCELO G. S.. Cooperative Parameter Estimation on the Unit Sphere Using a Network of Diffusion Particle Filters. IEEE SIGNAL PROCESSING LETTERS, v. 27, p. 715-719, 2020. (18/26191-0)