Improvement of RTT Fairness Problem in BBR Congestion Control Algorithm by Gamma Correction

Area of Science:

• Computer Science
• Network Engineering
• Data Communication

Background:

• Google's Bottleneck Bandwidth and Round-Trip Propagation Time (BBR) is a novel congestion control algorithm.
• BBR models network paths using bottleneck bandwidth and minimum Round-Trip Time (RTT) to optimize delivery rate and latency.
• Existing research highlights significant RTT fairness issues in BBR, where longer RTT flows dominate bandwidth, starving shorter RTT flows.

Purpose of the Study:

• To address the RTT fairness problem inherent in the BBR congestion control algorithm.
• To propose an improved algorithm, BBR-gamma correction (BBR-GC), that enhances fair bandwidth competition among flows with varying RTTs.

Main Methods:

• The proposed BBR-GC algorithm incorporates RTT as feedback information.
• A gamma correction function is employed to fit the adaptive pacing gain.
• Network performance was simulated using Network Simulator 3 (NS3).

Main Results:

• BBR-GC effectively alleviates the RTT fairness problem in BBR flows across different network conditions.
• The algorithm maintains high channel utilization.
• Simulations demonstrated a 50% improvement in RTT fairness and over 26% reduction in retransmission rates compared to original BBR.

Conclusions:

• The BBR-GC algorithm successfully resolves the RTT fairness issues associated with BBR.
• This improvement leads to more equitable bandwidth distribution among competing network flows.
• BBR-GC offers enhanced network performance by improving fairness and reducing retransmissions.