Mobile edge computing (MEC) can reduce the latency of cloud computing successfully. However, the edge server may fail due to the hardware of software issues. When the edge server failure happens, the users who offload tasks to this server will be affected. How to recover the services for these affected users quickly and effectively is challenging. Moreover, considering that the server failure is continuous and temporary, and the failed server can be repaired, the previous works cannot handle this problem effectively. Therefore, in this paper, we propose the fast, online, distributed, and temporary failure recovery algorithm (FODT) for MEC. In FODT, when edge sever failure happens, only the affected APs recalculate their user-server allocation strategies and the other APs do not change their strategies. For the affected access points (Aps), the strategies before server failure are reused to reduce complexity and latency. When the failed server is repaired, the influenced APs reuse the strategies before server failure to offload task to this server. Based on this approach, the FODT can achieve better performance than previous works. To the best of knowledge, the FODT is the first failure recovery algorithm, and when compared with previous research, it has higher failure recovery efficiency and lower complexity with acceptable approximate ratio.

翻译：移动边缘计算（MEC）可有效降低云计算延迟。然而，边缘服务器可能因硬件或软件问题发生故障。当边缘服务器故障发生时，将任务卸载至该服务器的用户将受到影响。如何快速有效地为这些受影响用户恢复服务极具挑战性。此外，考虑到服务器故障具有连续性与临时性特征，且故障服务器可被修复，现有方法难以有效处理该问题。因此，本文提出面向MEC的快速、在线、分布式与临时性故障恢复算法（FODT）。在FODT中，当边缘服务器发生故障时，仅受影响的接入点（AP）重新计算用户-服务器分配策略，其他AP保持不变。对于受影响AP，复用故障发生前的策略以降低复杂度与延迟。当故障服务器修复后，受影响AP重新使用故障前策略将任务卸载至该服务器。基于此方法，FODT能够实现优于现有方法的性能。据我们所知，FODT是首个故障恢复算法，与现有研究相比，在可接受的近似比条件下具有更高的故障恢复效率与更低的复杂度。