Optimization of Partially Offloading Mobile User Tasks to Fog Computing Networks

Abstract

Due to the limited computing power and energy resources of lightweight mobile user devices, offloading their tasks to fog nodes is a promising solution. Partial offloading is a more efficient way to allocate and use the whole system's resources than binary offloading. In this article, three algorithms are proposed to address the three issues of partial offloading in fog computing. Different from the strategy of minimizing task delay or minimizing power consumption adopted in the existing works of literature, the first algorithm considers both of these two strategies concurrently. The second is proposed to maximize task admission under the constraints of the user device power and system resource. The third is proposed to expand the system resource with the least cost to improve the admission when the system resource is insufficient. The above three are non-deterministic polynomial-time (NP)-hard problems, which will be solved by adopting the strategy of rounding algorithm in this article. Through theoretical analysis, a small gap exists between the solved solution and the optimal one of the problem. Additionally, a large number of simulations will be carried out to verify that their performance is better than that of the recent related works and close to the best performance.