TCP BBR congestion control optimization algorithm for high fairness and low retransmission

Abstract

Abstract Google proposed a new TCP congestion control algorithm based on Bottleneck Bandwidth and Round-trip propagation time (BBR). BBR achieves high transmission rate and low latency by estimating the available bottleneck capacity. Compared with the traditional loss-based congestion control algorithms such as CUBIC, BBR has obvious advantages in throughput. However, some studies have found that BBR has poor fairness among flows with different Round Trip Times (RTTs), inter-protocol unfairness and high packet loss rate in shallow buffer. BBR is already used in some cases, so its fairness and practicability improvements become very important. In this paper, the fairness of BBR is taken as the breakthrough point to deeply analyze the bandwidth probe mechanism. The pacing rate control model is derived and the effect of pacing gain on BBR performance is revealed. Furthermore, This paper proposes a BBR-PG algorithm, which adjusts pacing gain based on RTT to improve BBR. Simulation and real network experiments show that the proposed algorithm improves RTT fairness and inter-protocol fairness, reduces the retransmission rate,while retaining the throughput advantage of the original BBR.