Strategic Edge Cloud Placement via Clustering of Telecommunication Infrastructure

With the imminent increase in Internet users in Australia, where the average annual growth rate is 2.97%, this article addresses the growing demand for low-latency connectivity by optimizing mobile edge computing (MEC) placement using unsupervised clustering on Australian telecommunication site data. So, low-latency Internet is available to the maximum number of users as fast as possible. This paper explores finding the best locations to setup Mobile Edge Computing (MEC) servers, with the goal of improving internet speed and reducing response time for users across the country. Four clustering models-Gaussian Mixture Models (GMM), DBSCAN, Spectral Clustering, and a graph-based approach were applied to identify spatial patterns in infrastructure distribution. GMM revealed a well-balanced structure in ten clusters, the largest containing more than 17,000 sites. DBSCAN exposed a single dominant cluster of 95,404 sites, along with small dense pockets and noise points, indicating urban concentration and rural sparsity. Spectral clustering produced nuanced segmentation, capturing metropolitan and regional variations, with top clusters ranging from 19,985 to a few dozen sites. Cluster centroids were used to propose optimal edge server locations, balancing latency reduction with infrastructural feasibility. The results demonstrate strong clustering in both metropolitan areas and regional areas, reflecting diverse demands and deployment needs of the network. The findings provide a strategic framework for cost-effective MEC rollouts, enhancing service accessibility and responsiveness in dynamic telecom ecosystems.