Cost-Aware Placement Optimization of Edge Servers for IoT Services in Wireless Metropolitan Area Networks

Edge computing migrates cloud computing capacity to the edge of the network to reduce latency caused by congestion and long propagation distance of the core network. And the Internet of things (IoT) service requests with large data traffic submitted by users need to be processed quickly by corresponding edge servers. The closer the edge computing resources are to the user network access point, the better the user experience can be improved. On the other hand, the closer the edge server is to users, the fewer users will access simultaneously, and the utilization efficiency of nodes will be reduced. The capital investment cost is limited for edge resource providers, so the deployment of edge servers needs to consider the trade-off between user experience and capital investment cost. In our study, for edge server deployment problems, we summarize three critical issues: edge location, user association, and capacity at edge locations through the research and analysis of edge resource allocation in a real edge computing environment. For these issues, this study considers the user distribution density (load density), determines a reasonable deployment location of edge servers, and deploys an appropriate number of edge computing nodes in this location to improve resource utilization and minimize the deployment cost of edge servers. Based on the objective minimization function of construction cost and total access delay cost, we formulate the edge server placement as a mixed-integer nonlinear programming problem (MINP) and then propose an edge server deployment optimization algorithm to seek the optimal solution (named Benders_SD). Extensive simulations and comparisons with the other three existing deployment methods show that our proposed method achieved an intended performance. It not only meets the low latency requirements of users but also reduces the deployment cost.