QUIC Congestion Control Using RTT Feedback
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The QUIC protocol experiences inefficiencies in high bandwidth and high round trip delay environments due to its cubic congestion control algorithm, which inaccurately judges network status and limits the congestion window, leading to suboptimal throughput performance, especially under high bit error rates.
Innovation Solution
A congestion improvement method that updates the smooth round trip delay and self-adaptively adjusts the congestion window using the difference between expected and actual throughput, along with the minimum RTT, to accurately assess network conditions and prevent excessive packet loss, allowing for dynamic growth and reduction of the congestion window based on packet loss events.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If the cubic congestion control algorithm is used in QUIC protocol, then the throughput rate can be quickly improved when packet loss happens, but the algorithm cannot accurately judge the current network status and the congestion window cannot reflect the network status in time under high bandwidth and high round trip delay environment
Solution Approach 1:
The patent introduces a feedback mechanism by continuously monitoring RTT (Round Trip Time) and using it to dynamically adjust the congestion window. The system calculates expected throughput based on RTT measurements and compares it with actual throughput, using this feedback loop to accurately reflect current network status and make timely adjustments to the congestion window, resolving the issue of delayed network status reflection.
Solution Approach 2:
The patent changes the key parameter from static cubic-based window adjustment to dynamic RTT-based adjustment. By using RTT as the primary parameter for congestion window calculation, the system adapts to varying network conditions in real-time, improving measurement precision of network status while maintaining high throughput rates through parameter-driven dynamic adjustment.
2Speed
If the cubic congestion control algorithm is used, then the congestion window can quickly reach the target value after packet loss, but the congestion window will reduce sharply under high bit error rate environment
Solution Approach 1:
The patent makes the congestion window dynamic by continuously adjusting it based on current RTT measurements rather than using fixed cubic growth curves. This dynamic adjustment allows the system to quickly recover from packet loss when network conditions permit while providing stability when bit error rates are high, as the RTT-based mechanism naturally adapts to adverse conditions by reducing the window more gradually and responsively.
Solution Approach 2:
The system uses RTT measurements from the network itself to self-regulate the congestion window without external control. This self-service mechanism allows the protocol to automatically adapt to changing network conditions, including high bit error rates, by using the network's own response time characteristics to guide window adjustments, thereby maintaining both speed and reliability.
3Reliability
If the maximum congestion window is limited to 200 packets, then severe packet loss can be prevented, but the transmission performance is significantly affected in high bandwidth environment
Solution Approach 1:
The patent replaces the static maximum window limit of 200 packets with a dynamic window size determined by RTT-based calculations. This allows the congestion window to expand to utilize available bandwidth in high-speed networks while the continuous RTT monitoring ensures that the window is reduced promptly when packet loss occurs, thus maintaining reliability without sacrificing transmission performance.
Solution Approach 2:
The patent changes the fixed parameter (maximum window of 200) to a variable parameter that adapts to network conditions. By using RTT measurements to calculate the appropriate congestion window size, the system can achieve larger windows in high-bandwidth environments to improve throughput while maintaining packet loss prevention through responsive RTT-based adjustments.
4Productivity
If RTT-based congestion control methods like TCP-Westwood or TCP-Vegas are used, then transmission performance can be improved under long round trip delay, but the evaluation on RTT is affected by noises and larger difference appears
Solution Approach 1:
The patent applies smoothing techniques to RTT measurements to cushion against the effects of noise and variability. By using smoothed RTT values rather than raw measurements, the system reduces the impact of transient noise while maintaining the ability to respond to genuine network condition changes, thus improving measurement precision without sacrificing transmission performance.
Solution Approach 2:
The patent uses a feedback mechanism that continuously monitors RTT and adjusts the congestion window based on smoothed RTT values. This feedback approach allows the system to filter out noise through smoothing while still responding appropriately to real network conditions, resolving the contradiction between utilizing RTT for performance improvement and dealing with RTT measurement inaccuracies.
Data Source
AI summary
A kind of congestion improvement method based on the QUIC protocol adds the information of round trip delay in the congestion algorithm, self-adaptive changes the value of α to judge the situation of current network through comparison between the RTT of last time and the current RTT and then adjusts the current target window value in accordance with the current network situation, changing the congestion window based on the cubic growth curve of the cubic algorithm. This improvement method can make the QUIC protocol judge the current network situation more timely and accurately and can make the congestion window change quickly to fully utilize the bandwidth. The maximum congestion window limitation 200 exists in the QUIC protocol, which will not exceed 200 no matter how the congestion window grows. Such limitation largely reduces the throughput rate of QUIC protocol in the network environment with high bandwidth and long round trip delay.