Business‐driven policy optimization for service management

Abstract

The performance of services offered by network operators has a direct impact on its reputation, on its revenue due to new customer subscriptions, and also on penalties that can apply when services are not provided to an acceptable quality level. Previous research on business-oriented network and service optimization has mainly focused on optimizing individual business indicators, such as profit and revenue, in isolation without analyzing the effect on network configurations and the subsequent impact on other indicators. Given that different business objectives are usually incompatible, a single network configuration cannot optimize them simultaneously. Determining the configuration and the associated trade-offs that satisfy multiple objectives is a complex task. This paper addresses this gap and presents a framework that derives policy configurations that optimize the business value of the network infrastructure. We describe a methodology to quantify\ud business functions considering the dynamics of network events, the dynamics of end-user service usage, the nature of the business indicators, and their relationships with the underlying control methods. The proposed approach addresses the complexity of the target problem through a surrogate-based optimization approach properly tailored to match our application domain needs. We evaluate the effectiveness of the proposed approach through experimentation in a simulation environment we developed over OPNETPeer ReviewedPostprint (author’s final draft