The Decentralized Service based on Spatial Clustering

The development of internet services and the increased numbers of clients lead to develop new techniques that enhance the performance of the networks and reduces the latency. The request of the client goes to the data center in order to be served. When there are many clients those are connected to the data center at the same time some the requests will be delayed till the server finished the current requests. Therefore, different methods are deployed to reduce the latency. One of these methods is the decentralization of the service in which one or more of the nodes of the system, are used as spatial distributed nodes to manage the requests especially if the requests demands are migrated data at the same network. In this paper, A spatial distributed nodes are clustered by K-mean clustering algorithm to facilitate the selection of the decentralization nodes. For the given site of Mosul university in Iraq, the locations of the nodes are determined according to the buildings and establishments of the university. The nodes are clustered by K-mean to 5 clusters as an optimal number of clusters without outliers. The data center is located at the center of the system site. For each cluster, the nearest node to the data center is recognized as decentralization nodes. The centroids of the clusters are considered as Migration nods. The results showed that the for each node, the latency will be reduced. For 50, 100, 250, 500 clients, the latency is reduced to 40% by implementing decentralization with spatial clustering. © 2025 Elsevier B.V., All rights reserved.