BBR Fairness Evaluation Using NS-3

TL;DR

This paper assesses the fairness of BBR congestion control in network simulations, revealing unfairness issues when multiple BBR flows compete, influenced by network parameters like RTT and buffer size.

Contribution

It provides a detailed evaluation of BBR fairness using NS-3, highlighting the causes of unfairness and the need for mechanisms to improve equitable bandwidth sharing.

Findings

Unfairness increases with RTT and buffer size.

BBR outperforms loss-based methods in single flow scenarios.

Lack of response mechanisms causes convergence issues.

Abstract

This paper evaluates the fairness of BBR congestion control using NS-3 simulator. While BBR improves performance over loss-based methods in single flows, unfairness issues emerge with competing BBR and BBR/Cubic flows. Unfairness correlates with factors like round-trip time and buffer size. The core reason is the lack of responding mechanisms for the flows to converge on fair bandwidth share.