B1464
Title: Quadratic filter for networked systems with random parameter matrices and correlated noises under deception attacks
Authors: Raquel Caballero-Aguila - Universidad de Jaen (Spain) [presenting]
Josefa Linares-Perez - Universidad de Granada (Spain)
Abstract: In the context of networked systems, different random uncertainties usually degrade the performance of least-squares (LS) linear estimation algorithms. As a result, considerable efforts have been devoted to finding new types of suboptimal estimators. Among them, LS quadratic estimators have attracted the interest of researchers due to their balance between computational complexity and estimation accuracy. The goal is to address the LS quadratic filtering problem under the premise that the measurements are affected by random parameter matrices and correlated additive noises. The use of random parameter matrices models a broad variety of common uncertainties and random failures and thus better reflects engineering reality. In addition, the measured outputs are assumed to be vulnerable to malicious deception attacks and Bernoulli random variables are considered to depict the fact that such attacks occur randomly. By stacking the original vectors with their second-order Kronecker powers, the signal and observation vectors are augmented; then, using the rules of Kronecker algebra and an innovation approach, the linear estimator of the original signal based on the augmented observations is obtained, providing the required quadratic estimator. A simulation example shows how the designed quadratic filter outperforms the standard linear filter and how deception attacks affect the estimation performance.
