Fairness Analysis and Optimization of BBR Congestion Control Algorithm

Abstract

In distributed computing, data transmission is a crucial component, and the efficiency and fairness of transmission are key factors influencing system performance. Bottleneck bandwidth and round-trip time (BBR) is a new congestion control algorithm, it can improve the efficiency and stability of data transmission, thereby enhancing the overall performance of distributed computing. It aims to improve the performance of traditional Transmission Control Protocol (TCP) by measuring the bottleneck bandwidth and round-trip propagation time. However, in many cases, the BBR algorithm suffers from serious fairness problems, mainly in terms of its internal round-trip time (RTT) fairness. We construct a simple network topology based on NS-3 to evaluate the fairness problem. To optimize the RTT fairness problem of BBR, we propose the BBR-Optimization (BBR-O) algorithm, its pacing_gain is related to the size of the RTT. The experimental results show that the BBR algorithm prefers long RTT flows. In contrast, the BBR-O algorithm can effectively reduce the goodput difference between flows with different RTT sizes, increasing the values of inflight and sendrate for short RTT flows. The BBR-O algorithm balances the sendrate between different flows by setting the pacing_gain concerning the RTT size, which effectively alleviates the RTT fairness problem of BBR.