CMStatistics 2020: Start Registration
View Submission - CMStatistics
Title: Centralized estimation from measurements randomly subject to deception attacks Authors:  Raquel Caballero-Aguila - Universidad de Jaen (Spain) [presenting]
Aurora Hermoso-Carazo - Universidad de Granada (Spain)
Josefa Linares-Perez - Universidad de Granada (Spain)
Abstract: Cyber-attacks are becoming one of the most popular deliberate ways to reduce the reliability of a network. A typical kind of cyber-attacks is the so-called deception attack, which may include a wrong sensor measurement or control input, an incorrect time-stamp or a wrong identity of the sending device. The focus is on the least-squares linear centralized estimation problem for discrete-time stochastic signals from measured outputs provided by different sensors, which transmit their observations to a processing center where the estimator is designed. Deception attacks to the sensor network are assumed to be launched by an adversary and the success probabilities of these attacks, which may be different for each sensor, are known. The false information sent by the adversary involves two components: one that neutralizes the actual measurements and a noise component, which is the added blurred information. Hence, at each sampling time, the processing center may receive from each sensor either the actual measurement or just the noise injected by the adversary. Using the information received and by an innovation approach, a recursive centralized linear filtering algorithm is obtained without requiring full knowledge of the signal evolution model, but only the first- and second-order moments of the processes involved in the observation equations.