Optimization of collaborative resource allocation for mobile edge computing

Abstract

In order to realize the collaborative resource allocation optimization of mobile edge computing (MEC) and reduce the delay of edge server in the transmission process, based on software-defined network (SDN) technology, two optimal edge server deployment schemes of Enumeration-Based Optimal Edge Server Placement Algorithm (EOESPA) and Ranking-based Near-optimal Edge Server Placement Algorithm (RNOESPA) are proposed. Performance comparison experiment simulation is conducted with K-Means cluster algorithm (KMCA) to verify the minimum access delay of edge server under different conditions. After the deployment of edge servers, three collaborative resource allocation optimization algorithms of Optimal Enumeration Service Deployment Algorithm (OESDA), Latency Aware Heuristic Service Deployment Algorithm (LAHSDA), and Clustering Enhanced Heuristic Service Deployment Algorithm (CEHSDA) are proposed, and simulation experiments are carried out to verify the performance of the proposed algorithm under different conditions. The results show that, under different conditions, when the number of deployments increases from 1 to 4, the average access delay of EOESPA can be at least 1ms, and the average access delay obtained by RNOESPA is close to the best performance obtained by EOESPA and better than that obtained by KMCA. When the number of network nodes increases to 50, the minimum average access delay obtained by RNOESPA is closer to the optimal value, which is about 1.42ms. The same performance is shown in relation to the average number of requests, the number of mobile devices, and the average access delay. Among the three collaborative resource allocation optimization algorithms, the minimum average response delay obtained by LAHSDA is close to the optimal average response delay obtained by OESDA, but all of them are lower than CEHSDA, and CEHSDA has the best performance in minimizing the total allocation cost. When the number of service types increases to 8, the total service configuration cost of CEHSDA is about 0.89. It can be concluded that by optimizing the deployment of the edge server, the collaborative optimal allocation of its resources can be realized.