TY - JOUR
T1 - Resilient reinforcement learning and robust output regulation under denial-of-service attacks
AU - Gao, Weinan
AU - Deng, Chao
AU - Jiang, Yi
AU - Jiang, Zhong Ping
N1 - Funding Information:
Prof. Jiang is a recipient of the prestigious Queen Elizabeth II Fellowship Award from the Australian Research Council, CAREER Award from the U.S. National Science Foundation, JSPS Invitation Fellowship from the Japan Society for the Promotion of Science, Distinguished Overseas Chinese Scholar Award from the NSF of China, and several best paper awards. He has served as Deputy Editor-in-Chief, Senior Editor and Associate Editor for numerous journals. Prof. Jiang is a Fellow of the IEEE, a Fellow of the IFAC, a Fellow of the CAA and is among the Clarivate Analytics Highly Cited Researchers. In 2021, he is elected as a foreign member of the Academia Europaea.
Funding Information:
This work has been supported in part by the U.S. National Science Foundation under Grant EPCN-1903781 and CMMI-2138206 , and the Scientific Starting Fund from Nanjing University of Posts and Telecommunications (NUPTSF) under Grant NY221007 . The material in this paper was not presented at any conference. This paper was recommended for publication in revised form by Associate Editor Kyriakos G. Vamvoudakis under the direction of Editor Miroslav Krstic.
Publisher Copyright:
© 2022 Elsevier Ltd
PY - 2022/8
Y1 - 2022/8
N2 - In this paper, we have proposed a novel resilient reinforcement learning approach for solving robust optimal output regulation problems of a class of partially linear systems under both dynamic uncertainties and denial-of-service attacks. Fundamentally different from existing works on reinforcement learning, the proposed approach rigorously analyzes both the resilience of closed-loop systems against attacks and the robustness against dynamic uncertainties. Moreover, we have proposed an original successive approximation approach, named hybrid iteration, to learn the robust optimal control policy, that converges faster than value iteration, and is independent of an initial admissible controller. Simulation results demonstrate the efficacy of the proposed approach.
AB - In this paper, we have proposed a novel resilient reinforcement learning approach for solving robust optimal output regulation problems of a class of partially linear systems under both dynamic uncertainties and denial-of-service attacks. Fundamentally different from existing works on reinforcement learning, the proposed approach rigorously analyzes both the resilience of closed-loop systems against attacks and the robustness against dynamic uncertainties. Moreover, we have proposed an original successive approximation approach, named hybrid iteration, to learn the robust optimal control policy, that converges faster than value iteration, and is independent of an initial admissible controller. Simulation results demonstrate the efficacy of the proposed approach.
KW - Denial-of-service attacks
KW - Hybrid iteration
KW - Reinforcement learning
KW - Robust output regulation
UR - http://www.scopus.com/inward/record.url?scp=85129963203&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85129963203&partnerID=8YFLogxK
U2 - 10.1016/j.automatica.2022.110366
DO - 10.1016/j.automatica.2022.110366
M3 - Article
AN - SCOPUS:85129963203
SN - 0005-1098
VL - 142
JO - Automatica
JF - Automatica
M1 - 110366
ER -