Application Based Energy Optimization for Computation Offloading in Hierarchical MEC Network

Multi-access edge computing (MEC) is an emerging solution for computation offloading issues of mobile applications that require very low latency and high computing capacity. Since MEC hosts can be deployed in a hierarchical network, energy efficiency in host selection and migration problems of computation offloading is a critical challenge for MEC service providers. In this paper, the energy optimization problem is formulated to minimize the total energy consumption of the MEC system by considering application requirements in terms of latency constraints and user mobility. In this paper, user mobility is divided into multiple timeslots where each timeslot represents the time a user spent under the coverage area of a MEC host. The shortest path problem is derived from the energy optimization problem considering all timeslots and application migrations. Dijkstra’s algorithm is employed to find the solutions. Results from extensive simulations show that despite the fact that it is energy-efficient to serve the higher computing-intensive applications in higher level (deeper) of the hierarchical network, user mobility pattern and maximum tolerable latency of the application are critical factors in deciding the best MEC host to serve the offloading request. The results that measure the quality of service of latency-critical applications in terms of tasks served percentage, can be used for MEC network resource planning.