An Impact of Different Uncertainties and Attacks on the Performance Metrics and Stability of Industrial Control System

Abstract

The revolutionized growth in the networked control system has made the system more vulnerable to different types of threats such as delays, packet losses, and attacks. Such uncertainties in the control system may be introduced in the forward and/or feedback direction by the intended attackers while transmitting the signal. Herein, the system is estimated by modeling different parameters along with the knowledge about the actual system while considering the delay, packet loss rate, etc., using system identification and True-Time tool. Also, the particle swarm optimization (PSO) algorithm is used for modeling the system and estimation of controller parameters. The efficacy of the estimated model is shown by evaluating the various performance measures along with the introduction of design of a model of attacks based on the estimated system parameters to compromise the actual network control performance. In the absence of attacks, the estimated model used here shows above 94% fit to the estimation data which indicate the efficacy of the estimated model. Later, different rates of packet loss such as 0, 5, 15, and 25% are introduced for estimating the system model. For each case, the attack is simulated to show the impact of service degradation attack on the performance metrics of the networked control system. The overshoots are 15.698% for 0% packet loss, 17.4% for 5% packet loss, 17.9% for 15% packet loss, and 17.5% for 25% packet loss along with modeled attack. The other metrics such as rise time, settling time, and peak time increase substantially, whereas root mean square value decreases, and the cost function increases for estimated system with different packet loss rate and attack, which is an indication of a compromised system.