Due to increasing complexity of mobile applications, and limited computation resources of smart mobile devices, the quality of service requirements of mobile application can be enhanced by offloading the computation tasks of the mobile applications to edge servers, such as cloudlets, which exist at the edge of wireless networks. However, improper placement of mobile tasks on the edge servers may increase the waiting time and the transmission time. This, in turn, will increase the response time, and eventually violates the quality of service.

This paper proposes an effective offloading strategy in a mobile edge environment using the queuing networks and an evolutionary algorithm, namely the genetic algorithm (GA). The queuing network is used to model the waiting time and the service time of the mobile tasks on the edge servers. The genetic algorithm finds the best allocation of mobile tasks on the edge servers to minimize tasks response time considering the transmission times and the load conditions on edge servers represented by the waiting times and the service times which are calculated using the queuing network. The proposed GA-based offloading algorithm is compared with another evolutionary algorithm, namely particle swarm optimization (PSO). Experimental evaluations show that the GA-based offloading algorithm outperforms both of round robin offloading and the PSO-based offloading algorithms, and effectively improves mobile applications response time.

由于移动应用程序的复杂性增加，智能移动设备的计算资源有限，移动应用程序的服务质量要求可以通过将移动应用程序的计算任务卸载到边缘服务器（例如存在于边缘的小云）上来提高的无线网络。然而，移动任务在边缘服务器上的不当放置可能会增加等待时间和传输时间。这反过来又会增加响应时间，并最终违反服务质量。

本文提出了一种在移动边缘环境中使用排队网络和进化算法的有效卸载策略，即遗传算法（GA）。排队网络用于对边缘服务器上移动任务的等待时间和服务时间进行建模。遗传算法在边缘服务器上寻找移动任务的最佳分配，以最小化任务响应时间，考虑边缘服务器上的传输时间和负载条件，由使用排队网络计算的等待时间和服务时间表示。将提出的基于 GA 的卸载算法与另一种进化算法进行比较，即粒子群优化 (PSO)。