Path selection with Nash Q‐learning for remote state estimation over multihop relay network

Abstract

SummaryIn this article, the security issue of remote state estimation is investigated for multihop relay networks interrupted by an attacker launching denial‐of‐service attacks. Since the presence of the relay enriches the communication topology, there might exist several paths connecting the sensor and the estimator, consisting of the corresponding channels. Thus, it is reasonable for the sensor to select the path with a lower dropout rate to enhance the system performance measured by the estimation error, due to the dropout rate changing with the channel. However, as an adversary, the objective of the jammer is to deteriorate the corresponding performance through launching attack on the communication path selectively. For addressing the problem on the behalf of both of the sensor and the jammer, we first formulate this problem as a two‐player zero‐sum stochastic game model, and then present a Nash Q‐learning algorithm to explore the equilibrium point for both players, under the assumption that both of them are rational players. Furthermore, the existence of equilibrium point for this problem is proved analytically. Moreover, a more general case of the channel attack, under which the jammer can attack any channels among this network, is considered. Finally, numerical results are presented to verify the effectiveness of the algorithm proposed and the theorem results.

A reawakening of Machine Learning Application in Unmanned Aerial Vehicle: Future Research Motivation

“…QL can be simply defined as a model-free RL process to acquire strategies expressively informing the agent on what action to do depending on a given situation. Some functions are returned to be the recompressions rummagesale to deliver the strengthening providing the quality of an action reserved in a specified situation [34] and [35].…”