Secure Estimation in V2X Networks with Injection and Packet Drop Attacks

Abstract

Vehicle-to-anything (V2X) communications are essential for facilitating cooperative intelligent transport system (C- ITS) components such as traffic safety and traffic efficiency applications. Integral to proper functioning of C- ITS systems is sensing and telemetery. To this end, this paper examines how to ensure security in sensing systems for V2X networks. In particular, secure remote estimation of a Gauss-Markov process based on measurements done by a set of vehicles is considered. The measurements are collected by the individual vehicles and are communicated via wireless links to the central fusion center. The system is attacked by malicious or compromised vehicles with the goal of increasing the estimation error. The attack is achieved by two mechanisms: false data injection (FDI) and garbage packet injection. This paper extends a previously proposed adaptive filtering algorithm for tackling FDI to accommodate both FDI and garbage packet injection, by filtering out malicious observations and thus enabling secure estimates. The efficacy of the proposed filter is demonstrated numerically.