Federated Learning Based Secured Computational Offloading in Cyber-Physical IoST Systems

In the past years, there has been fast growth in various applications which generates penetrating and personal information dependent on the Internet of things (IoT). The Softwarization of a network may improvise its inclusive flexibility. The software-enabled proficiency has made it functional for mobile devices, wireless, and telecommunication networks through the progression of technologies like software-defined network (SDN), and mobile edge networks. The applications of cyber-physical systems have also increased with the rise of cyber-physical Internet of softwarized things (IoST) systems. To maintain the quality among users, managing the applications for latency of cyber-physical system (CPS) is mandatory. Though edge computing and cloud computing have performed well in achieving latency-aware resource allocation, sometimes it fails to do secure computation offloading in cyber-physical IoST systems. This paper presents a novel federated scheme based on deep learning for fast computation offloading. Secondly, we develop a federated learning framework that allows cyber physical IoST systems to build a model in a privacy-preserving way.