Optimal Control for Networks with Unobservable MaliciousNodes
Bai Liu, Eytan Modiano · ACM SIGMETRICS Performance Evaluation Review · 2022
With the rapid development of information technology, modern network systems are becoming increasingly complex and are increasingly vulnerable to attacks such as Distributed Denial-of-Service (DDoS) attack. However, existing network control algorithms either require full observability and/or controllability for all nodes [2, 9, 1, 4, 6], or the network dynamics to be time-invariant and stochastic [7, 3, 8, 5]. In this paper, we aim to develop a new algorithm that can stabilize networks with unobservable and uncontrollable nodes under adversarial dynamics (i.e., external arrivals and actions of malicious nodes).