Delay Attack and Detection in Feedback Linearized Control Systems

Delay injection attacks on nonlinear control systems may trigger instability mechanisms like finite escape time dynamics. The paper guards against such attacks by showing how a recursive algorithm for identification of nonlinear dynamics and delay can simultaneously provide parameter es-timates for controller tuning and detection of delay injection in the feedback path. The attack methodology is illustrated using a simulated feedback linearized automotive cruise controller where the attack is disguised, but anyway rapidly detected.