Improving prediction accuracy of Matrix Factorization based Network coordinate systems

Abstract

Matrix factorization (MF) based Network coordinate (NC) systems solve the triangle inequality violations (TIVs) that is the main problem of Euclidean distances. However, these systems suffer from low prediction accuracy. In this paper, Conditional Clustered Network Coordinate (CCNC) System is proposed. It divides the space into a number of clusters in a balanced, dynamic, and decentralized way. Clustering the whole space is based on two thresholds in order to guarantee a balanced clustered operation. The performance of CCNC system is evaluated with King data set and PlanetLab data set to be compared against two well known NC systems: Phoenix and Pancake. The simulation results show that CCNC outperforms Phoenix and Pancake significantly in terms of estimation accuracy, expected time to construct the clusters, and the communication overhead.