Service Placement Considering Robustness and Dynamic in Edge Computing

Abstract

With the increasing popularity of the Internet of Things, a huge requirement for low-latency in service response has spurred. Many researches have been proposed on service placement in mobile edge computing (MEC) to address the low-latency requirement, which pushes more service functions from the remote cloud to the network edge. However, they do not consider the scenarios with uncontrolled factors, such as disaster relief and battlefield monitoring, the edge server is likely to go down and fail to respond to the service requests from users. In this paper, we defined the service placement problem in the unreliable and dynamic network and developed a novel distributed service placement decision-making solution called RTSO which based on the greedy algorithm. The edge server using the RTSO algorithm will take into account service popularity and resource constraints, and judiciously adjust service placement decisions. We also verify through experiments that our proposed algorithm can converge to the global optimum through iteration. A large number of simulation results show that our algorithm can provide near-optimal performance and effectively improve the robustness of services in the system.