Date of this Version
Shreyas Sundaram, Jian Chang, Krishna K. Venkatasubramanian, Chinwendu Enyioha, Insup Lee, and George Pappas, "Reputation-Based Networked Control With Data-Corrupting Channels", 14th International Conference on Hybrid Systems: Computation and Control (HSCC '11) , 291-300. April 2011. http://dx.doi.org/10.1145/1967701.1967743
We examine the problem of reliable networked control when the communication channel between the controller and the actuator periodically drops packets and is faulty i.e., corrupts/alters data. We first examine the use of a standard triple modular redundancy scheme (where the control input is sent via three independent channels) with majority voting to achieve mean square stability. While such a scheme is able to tolerate a single faulty channel when there are no packet drops, we show that the presence of lossy channels prevents a simple majority-voting approach from stabilizing the system. Moreover, the number of redundant channels that are required in order to maintain stability under majority voting increases with the probability of packet drops. We then propose the use of a reputation management scheme to overcome this problem, where each channel is assigned a reputation score that predicts its potential accuracy based on its past behavior. The reputation system builds on the majority voting scheme and improves the overall probability of applying correct (stabilizing) inputs to the system. Finally, we provide analytical conditions on the probabilities of packet drops and corrupted control inputs under which mean square stability can be maintained, generalizing existing results on stabilization under packet drops.
CPS Embedded Control
14th International Conference on Hybrid Systems: Computation and Control (HSCC '11)
© ACM 2011. This is the author's version of the work. It is posted here for your personal use. Not for redistribution. The definitive Version of Record was published in Proceedings of the 14th International Conference on Hybrid Systems: Computation and Control (HSCC '11), http://dx.doi.org/10.1145/1967701.1967743.
Reliability, Security, Design, Algorithms, Theory, Networked Control, Reputation, Majority Voting, Stability Analysis
Date Posted: 03 May 2011
This document has been peer reviewed.