Energy-Efficient Task Offloading for Time-Sensitive Applications in Fog Computing

Abstract

Fog computing has attracted considerable attention to meet the location awareness and real-time response requirements of various applications. However, compared to cloud computing, fog devices have relatively limited power supplies, computation resources, and communication resources, raising design challenges to meet real-time response requirements. To balance response time and energy consumption for multiple fog devices running multiple applications, this work presents an energy-efficient offloading decision mechanism and an offloading dispatcher to dispatch applications to the corresponding device by effectively managing the computation and communication resources of all devices in a fog computing environment. The offloading tasks are composed of several subtasks with an end-to-end deadline. To meet the response time requirements of such applications, a run-time scheduler with an end-to-end latency schedulability consideration is also presented. Evaluation results show considerable energy savings using this framework and a real platform study provides validation.