A Graph-Based Decision-Making Strategy for Cold Server Set Selection in Mobile Edge Computing

Le résumé fourni par la source

The purpose of mobile edge computing is to distribute various concurrent applications to multiple edge servers for execution. During the distribution process, some servers are only allocated a small number of functions, resulting in a waste of server energy consumption, and these servers are defined as “cold”. Such a defect can be solved by a migration algorithm, i.e., migrating the functions allocated to cold servers to other servers, and then shutting down the cold servers to reduce energy consumption. This article ignores the migration algorithm and focuses on the selection of cold servers. The cost of the path is defined by the summation of the costs of each edge on the path, and then the cheapest path is further introduced. The average difference function is designed to characterize the impact of deleting a specific cold server set with respect to the cheapest path of the remaining subgraph, and the optimal cold server set is determined in terms of our proposed decision making strategy. Furthermore, we propose a channel resource adjustment scheme in light of edge curvature calculation, which is used to retrieve the edge set that causes the bottleneck in the subgraph after deleting the cold server set. The proposed graph-based decision-making strategy consistently achieves a lower average distance compared to the greedy-based approach, max degree-based approach and random-based approach across various random connected weighted graphs and scale-free random graphs. This demonstrates its superior effectiveness in minimizing structural damage to the network when selecting cold server sets.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.