An Analytical Approach to Inventory Management for Telecommunications Network Equipment

Abstract

Verizon’s core network relies on routers that use expensive plug-in cards (PICs) to function. PIC failures can cause disruptions to FiOS or wireless service in the surrounding area. Maintaining an appropriate level of spare PICs is important to maintaining high service levels. Due to the high cost and volume of PICs used in Verizon's networks, it becomes expensive to hold excessive inventory onsite, without exponentially increasing working capital.

This project takes two approaches to aiding in inventory management: first, building predictive models for PICs failures in FiOS routers based on previously recorded failures; second, developing a holistic Central Office (CO, where telecommunications data is routed) prioritization system to determine where to place constrained inventory. While developing predictive models for PICs failures proved unsuccessful, the CO prioritization approach resulted in reductions in outage customer impact and traffic impact of more than 40% across all test cases.