| Abstract |
An edge computing system is an emergent architecture for providing computing, storage, control, and networking abilities, that is an important technology to realize Internet of Things and smart cities. In an edge computing environment, users can offload their computationally expensive tasks to offloading points, which may reduce the energy consumption or communication delay. There are a large number of offloading points and users in a system, and their tasks are structured. However, resources of offloading points are limited, and users have different preferences for energy consumption and communication delays. In this paper, we first establish a system model for the environment with multiple users, multiple offloading points, and structured tasks. Then, we formalize an offloading decision problem in such an environment as a cost-minimization problem, which is a NP-hard problem. Thus, we design a method based on backtracking to obtain its exact solution; the method's time complexity is, unfortunately, exponential with the number of offloading points. To reduce the complexity, a method based on an improved genetic algorithm and a method based on a greedy strategy are designed. Finally, we validate and compare three methods in terms of the total cost of all users, resource utilization of offloading points and execution time. The simulation results show that the last method performs the best. (C) 2020 Elsevier B.V. All rights reserved. |