Processing Delay Optimization for Data Intensive Applications for Fog/Cloud Computing Using Double Reinforcement Learning

With smartphones becoming our daily companions, it has generated high-quality mobile services everywhere and there is tremendous growth of mobile traffics. This is going to becomea huge hindrance to the next generation mobile networks. Hence fog computing with cloud computing to deal with highly data-oriented applications sounds like a practical solution. Reinforcement learning is sub branch of machine learning which offers multiple solutions and the learning agent makes decision about superlative solution. A novel Double Reinforcement Learning based delay optimization framework is designed to optimize the processed data which need to be stored to fog/cloud storage system. The agent consider dual Q-states to formulate optimal policies for optimization of processing delay encountered by High Performance Computing (HPC) applications. The agent performs very well under varying uncertainty of large state space computing system like cloud/fog. The performance of the proposed framework is good pertaining to performance metrics like processing delay, learning rate, throughput, and cost of operation.