Security Reassessing in UAV-Assisted Cyber-Physical Systems Based on Federated Learning

Abstract

Mobile-edge computing (MEC) is a popular method for increasing the quality of computing experience for consumers since it allows them to offload computing operations to MEC servers, which have significant processing capabilities. However, safety and information security are major challenges that must be addressed. We employ Federated Learning-based Cyber-Physical Systems (CPS) to address safety and information security in MEC systems, which offers data accuracy and un-sustainability. Security of UAV assisted CPS against cyber-attacks yet an another challenging problem. Because most cyber-attacks occur in unpredictable ways, it is difficult to define them in a structured manner. Instead of developing a unique cyber-attack model, we focus on exploring the dynamic behavior of the system to cyber-attacks throughout this article. Attacks that repeat themselves by interfering with system components or data are insignificant if they can be quickly identified by the system's control mechanism. Intelligent cyber attackers remain undetected by the tracking system by carefully designing cyber-attacks. Our primary goal is to examine the effectiveness of such cyberattacks from the aspect of the CPS and overcome them using Federated Learning (FL).