Analysis of the Cost of Varying Levels of User Perceived Quality for Internet Access

TL;DR

This paper analyzes how different levels of user perceived quality (QoE) for internet access relate to bandwidth costs, using MOS ratings to quantify improvements and providing insights for bandwidth management.

Contribution

It offers a detailed analysis of QoE versus bandwidth costs in internet access, especially considering TCP multiplexing, which is a novel extension beyond existing QoS-QoE mappings.

Findings

Increasing bandwidth improves MOS scores.

TCP multiplexing affects the QoE-cost relationship.

Provides practical guidance for bandwidth investment decisions.

Abstract

Quality of Service (QoS) metrics deal with network quantities, e.g. latency and loss, whereas Quality of Experience (QoE) provides a proxy metric for end-user experience. Many papers in the literature have proposed mappings between various QoS metrics and QoE. This paper goes further in providing analysis for QoE versus bandwidth cost. We measure QoE using the widely accepted Mean Opinion Score (MOS) rating. Our results naturally show that increasing bandwidth increases MOS. However, we extend this understanding by providing analysis for internet access scenarios, using TCP, and varying the number of TCP sources multiplexed together. For these target scenarios our analysis indicates what MOS increase you get by further expenditure on bandwidth. We anticipate that this will be of considerable value to commercial organizations responsible for bandwidth purchase and allocation.