A Predictive Approach for Managing Network Port Resources of Service Providers

Abstract

Internet service is highly dependent on the available network capacity (or network port resources) within the operation region of interest. Appropriate management of network port resources to provide Internet services for all subscribers (including new users) is a critical operation and need to be planned proactively for service providers. For this reason, prediction of available number of network port resources can provide proactive actions while ensuring enhanced decision making for operational units of service providers. This paper is investigating a prediction methodology to obtain the availability of number of network port resources for broadband IP network services using realworld case study of a major telecommunication infrastructure provider's data in Turkey. Our analysis is also considering the practical connection ranges of each region types (e.g. hospitals, airports, industrial, etc). Our predictor variables include both in-network based data (such as past and current total number of subscribers, number of available network ports) as well as outnetwork based data (such as population, online number of ads for sales and rental houses from a well-known Turkish advertisement website in the region of interest). Our results indicate that depending on the region of interest, prediction results on stable regions can yield better insights for service providers compared to booming regions of the country.