Self-Feedback-Based Resilient Consensus Network

There has been a wide interest in understanding the vulnerabilities of consensus network that allows an attacker to cause harm. At the same time, it can be utilized to enhance resilience against such attackers. In this paper, we investigate vulnerabilities of a single integrator consensus network that can be exploited to launch an attack. We show that root nodes are more vulnerable as compared to non-root nodes. If an attack is injected on root nodes the network dynamics are destabilized, while, an attack on non-root nodes prevents agents from reaching consensus. Resilience against such attackers can be improved by adding self-feedback at each node of the network. We show that a self-feedback-based consensus network remains stable in the presence of a destabilizing attack. Moreover, we investigate the use of self-feedback at the root nodes and at the non-root nodes as well. We found that the placement of self-feedback only at root nodes is sufficient to ensure resilience against attack. Simulation examples are provided to validate the results developed in the paper.