SecOFF-FCIoT: Machine learning based secure offloading in Fog-Cloud of things for smart city applications

Computation offloading is one of the important application in Internet of Things (IoT) ecosystem. Computational offloading provides assisted means of processing large amounts of data generated by abundant IoT devices, speed up processing of intensive tasks and save battery life. In this paper, we propose a secure computation offloading scheme in Fog-Cloud-IoT environment (SecOFF-FCIoT). Using machine learning strategies, we accomplish efficient, secure offloading in Fog-IoT setting. In particular, we employ Neuro-Fuzzy Model to secure data at the smart gateway, then the IoT device selects an optimal Fog node to which it can offload its workload using Particle Swarm Optimization(PSO) via the smart gateway. If the fog node is not capable of handling the workload, it is forwarded to the cloud after being classified as either sensitive or non-sensitive. Sensitive data is maintained in private cloud. Whereas non-sensitive data is offloaded using dynamic offloading strategy. In PSO, the availability of fog node is computed using two metrics; i) Available Processing Capacity (APC), and ii) Remaining Node Energy (RNE). Selection of cloud is based on Reinforcement Learning. Our proposed approach is implemented for smart city applications using NS-3 simulator with JAVA Programming. We compare our proposed secure computation offloading model with previous approaches which include DTO-SO, FCFS, LOTEC, and CMS-ACO. Simulation results show that our proposed scheme minimizes latency as compared to selected benchmarks.