A Fully-Distributed Resilient Dynamic Output Feedback Consensus Control for MASs Under Markovian DoS Attacks

Muhammad Ahsen Ali, Muhammad Rehan, Naeem Iqbal, Abdul Basit, Choon Ki Ahn

Research output: Contribution to journalArticlepeer-review


This paper is concerned with the resilient leader-following consensus control (LFCC) of multi-agent systems (MASs) subjected to denial-of-service (DoS) attacks. Unlike most existing studies, the considered DoS attacks can independently jeopardize different communication links and are modeled using the Markovian switching phenomenon. First, a cooperative fully-distributed estimator is introduced that enables the follower agents to estimate the leader&#x2019;s state. The proposed cooperative estimator relaxes the requirement for knowledge about the exact leader state during the occurrence of DoS attacks. The cooperative estimator is then used to solve the resilient LFCC of MASs under two different control protocols. The dynamic state feedback (DSF)-based consensus protocol is first used to solve the fully-distributed resilient LFCC problem with cooperative estimation of the leader state. Then, to overcome the limitation of the follower&#x2019;s state unavailability, a novel dynamic output feedback (DOF)-based control protocol is introduced to solve the resilient consensus problem. The design conditions for the proposed estimators and controllers are derived using the Lyapunov theory. Unlike existing studies, the proposed strategy 1) employs estimated states for feedback, 2) is fully-distributed, and 3) is independent of bounds on attack frequency or duration. Finally, the applicability of the presented strategies is validated through a simulation example involving the F-18 aircraft model. <italic>Note to Practitioners</italic>&#x2014;The purpose of the present work is to provide an output feedback approach for the LFCC of MASs by incorporating cyber attacks. The cyber attacks in the form of DoS while communicating over a network have been considered. The state estimation problem over a network under DoS attacks on communication links has been solved by using the cooperative estimator equation. Then, the adaptive and smoothing coupling weights are updated by applying the local estimation error. The proposed approach employs an adaptive mechanism for handling DoS attacks over the communication topology to deal with unknown connections. This mechanism does not require knowledge of the topology between agents and, consequently, does not require the central unit for implementation. Then, two approaches are used to solve the problem of resilient cooperative control: 1) by employing the estimation of the leader&#x2019;s state only; 2) by employing both estimates of the leader&#x2019;s and followers&#x2019; states. The proposed approach can be applied practically by solving cooperative and conventional observers along with adaptation weights, recursively, in addition to an algebraic equation for estimated state feedback. Moreover, the presented method can deal with independent attacks on different communication channels, which is a practical scenario. The proposed approach can be applied to consensus and formation control applications under cyber attacks, like the formation of aircrafts, consensus in robots, and synchronization in power systems. The presented approach has been applied to attain consensus in F-18 aircrafts, and simulation results are shown.

Original languageEnglish
Pages (from-to)1-10
Number of pages10
JournalIEEE Transactions on Automation Science and Engineering
Publication statusAccepted/In press - 2024

Bibliographical note

Publisher Copyright:


  • denial-of-service attacks
  • Distributed algorithms
  • dynamic output feedback
  • Markovian switching topologies
  • multi-agent systems
  • resilient observer

ASJC Scopus subject areas

  • Control and Systems Engineering
  • Electrical and Electronic Engineering


Dive into the research topics of 'A Fully-Distributed Resilient Dynamic Output Feedback Consensus Control for MASs Under Markovian DoS Attacks'. Together they form a unique fingerprint.

Cite this