Double-State-Temporal Difference Learning for Resource Provisioning in Uncertain Fog Computing Environment

Abstract

The Internet of Things (IoT) has evolved at a faster rate in recent years. Fog computing can perform a substantial amount of computation at a faster rate and is capable of dealing with IoT applications with huge traffic demands and stringent Quality of Service (QoS) requirements. Resource provisioning issues in fog computing are addressed using double-state-temporal difference learning. By considering double Q-states while propagating the temporal difference error, high quality resource provisioning policies can be formed. The proposed resource-provisioning scheme outperforms one of the recent works in terms of the performance metrics such as execution time, learning rate, accuracy and resource utilization rate under varying uncertainties of the task and grid resource parameters.