Task Prediction and Optimal Offloading Decisions for Fog Computing Cloud Network
Dheeraj Sharma, Niraj Pratap Singh · 2024
Fog computing (FC) emerged as a model to facilitate delayed-sensitive tasks in mobile edge computing (MEC) devices and the Internet of Things (IoT) by providing alternative and shared computing and network resources, as well as cloud computing. FC is intended to deal with cloud-based networking&s;s lengthy latency and high connection load. The terminal node&s;s computational responsibilities may be offloaded to adjacent fog nodes, resulting in a substantially faster processing time than cloud-based networks. Current studies on FC networks have focused mainly on the overall energy used by executing a task. Nevertheless, particularly for battery-powered fog nodes, equitable offloading across numerous fog nodes while keeping low energy consumption and average response time is critical. For FC networks, this paper has presented an optimization-based algorithm, termed as long short-term memory—grey wolf optimization (LSTM-GWO), for task predicting and optimum offloading. The task delay is calculated, as well as the associated energy usage. Simulation results and comparison demonstrates that the suggested task offloading strategy performs well in FC networks considering average response time, load imbalance degree, and load imbalance standard deviation. The proposed LSTM-GWO task offloading surpasses existing ant colony optimization (ACO) and particle swarm optimization (PSO) task offloading in terms of average response time, load imbalance standard deviation, and degree of inequality (DI).