An efficient dynamic service provisioning mechanism in fog computing environment: A learning automata approach

Abstract

Recent advances in the Internet of Things (IoT) technology have contributed to growing the number of IoT applications in various scenarios, e.g., buildings, cities, healthcare, wearable devices, and businesses to change the way work and live. The fog computing model has appeared as an appropriate distributed platform to service at the edge of the network using the resource capacities to support and execute the real-time IoT applications. One of the most challenging issues in IoT application management is the dynamic service provisioning problem to fix changes in IoT application resource usage patterns. In this paper, we propose an efficient dynamic service provisioning mechanism using the learning automata technique to determine the service provisioning decisions to deploy or release the IoT applications over the heterogeneous and dynamic fog infrastructure. Besides, we design an autonomous dynamic service provisioning manager (DSPM) that follows a self-management control loop to provision the IoT applications on the fog infrastructure. The simulation results obtained using synthetic and real-world traffic traces show that our proposed algorithm effectively reduces service delay, cost and service delay violation compared to other algorithms.