ABSTRACT The Bottleneck Bandwidth and Round‐trip propagation time (BBR) protocol has recently been adopted on the Internet to achieve high sending rates, especially for multimedia applications. However, the gradual increase in sending rate overloads the bottleneck link, leading to persistent queues, higher queuing delay, increased packet loss, and overall network performance degradation. Further, intra‐protocol unfairness limits BBR's effectiveness in congestion control. In this paper, we address these limitations by proposing an adaptive pacing gain mechanism that dynamically responds to RTT variations. We formulate a multi‐variable optimization objective that jointly considers RTT variation, queuing delay, and pacing gain. The proposed BBR‐G is implemented in the ns‐3 network simulator and further validated using Mininet experimentation. The ns‐3 experiments are conducted across multiple topologies (many‐to‐one and many‐to‐many) and varied bottleneck bandwidth networks. The experimental results show that BBR‐G improves average network throughput up to 50.77%, reduces bottleneck queue occupancy up to 50.84%, and decreases delay up to 45.2% than BBR, BBR‐A, FEBBR, Vegas, and CUBIC. The ‐test‐based statistical validation of throughput alongwith ablation study and sensitivity analysis demonstrate the robustness of the proposed BBR‐G in simulation environments. Further, Mininet‐based emulation experiments are conducted on a shared bottleneck dumbbell topology with multiple concurrent flows to validate the practicality of the proposed approach. The experimental results demonstrate that BBR‐G achieves improved throughput and reduced latency while maintaining better fairness compared to existing congestion control algorithms.
S et al. (Thu,) studied this question.