Improving Task Migration Using Fuzzy Logic Type 2 and Reinforcement Learning Algorithm with Adjacent Policy to Enhance the Performance of IoT Applications

This paper presents a novel task migration method designed to improve the performance of IoT applications in a three-tier architecture in fog and cloud computing environments. The IoT layer consists of smart devices that generate a large number of tasks; each task has various characteristics such as size, computational requirements, communication requirements, and time constraints. Due to the storage and computational capacity limitations of IoT devices, these tasks need to be migrated to different layers to perform effective processing and meet Quality of Service (QoS) objectives. To solve this challenge, a type 2 fuzzy logic task scheduler is used to make intelligent decisions for task migration. This scheduler selects the most appropriate processing layer based on the task characteristics. In addition, in this paper, deep learning reinforcement learning neighbor policy optimization (PPO) is used to maintain load balance among peer nodes by appropriately migrating fog-level tasks. Experimental results show that the proposed scheme outperforms existing state-of-the-art methods in terms of latency reduction, energy consumption, network utilization, throughput, and task migration rate.