ANALYSIS OF NETWORK PROPERTIES USING SELF ORGANIZING MAPS FOR SERVICE DEPLOYMENT ON THE CLOUD

Abstract

This work offers in-depth analysis of network properties to employ them for service deployment on cloud systems. The proposed analysis is evaluated on three different data sets from different locations, captured in 2012, 2013 and 2014 to provide insights into network properties and how to use them while deploying services. This research proposes the employment of a Self-Organizing Map as a type of Artificial Neural Network that generates a low-dimensional representation of high dimensional data using unsupervised learning methods. My analysis shows that there are significant effects of selected network properties, namely latency, success status, hop count and time-to-live, on the optimal location for the service to be deployed. In summary, using the proposed technique for analysis of network properties to choose the location for service deployment on the cloud could help to understand where to deploy the service to increase efficiency with respect to the selected properties.