Noncooperative Formation Tracking of UAVs Against Multiple Cyber-Threats: A Twin-Network Approach

In this article, we investigate the problem of unmanned aerial vehicles (UAVs) formation tracking in the presence of multiple cyber-threats. In this scenario, UAVs have private, potentially conflicting objectives, and their communication networks and local feedback mechanisms are vulnerable to sabotage and eavesdropping by malicious attackers. To address this real-world challenge, we propose a control scheme based on a twin-network structure. Specifically, we construct a virtual twin layer interconnected with the physical layer to design a resilient estimator that fortifies information exchange among UAVs under threats. In addition, by coupling the states from the twin layer and time-varying signals as masks, we achieve privacy protection for critical information. Furthermore, leveraging reliable data provided by the resilient estimator, we design a cooperative controller based on gradient-optimization to update the UAVs' positions. Using Lyapunov theory, we prove that the position of all UAVs converge to a dynamic Nash equilibrium. Finally, we conduct experimental studies to validate the effectiveness of the proposed control scheme.

I. Introduction

Recent advancements in formation control of unmanned aerial vehicles (UAVs) have provided effective tools for addressing various civil and military tasks in the airspace, such as target tracking and localization [1], communication relay [2], and cooperative hunting [3]. It is increasingly recognized that achieving an ideal formation is often a prerequisite for UAVs collaborative missions. Consequently, representative formation control methods have garnered continuous attention from scholars.