Network Congestion Control via Active Flow Feedback

Solution Overview

Problem

Conventional congestion control methods in data center networks are inefficient due to slow rate adjustments, leading to high delay and low congestion control efficiency.

Innovation Solution

A congestion control method where a first network device receives an active flow quantity from a second network device, determines packet sending control information based on this quantity and the rated receiving bandwidth, and adjusts the data flow accordingly to ensure the actual receiving bandwidth reaches the rated bandwidth.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If the transmit end gradually adjusts the sending rate after learning congestion state, then the sending rate adapts to network conditions, but the adjustment is slow and causes transmission delay

Engineering Contradiction:

Improvecongestion adaptationVSAvoidtransmission delay

Core Design Contradiction:

Adaptability or versatilityVSLoss of time

Solution Approach 1:

The patent applies preliminary action by having the receive end proactively calculate and feed back the optimal sending rate to the transmit end before congestion occurs. The receive end determines the optimal rate based on its receiving capability and active flow quantity, then sends this information in advance via feedback messages, allowing the transmit end to adjust immediately without waiting for congestion to manifest.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent implements feedback by establishing a bidirectional communication mechanism where the receive end continuously monitors its receiving state, calculates optimal sending rates, and feeds this information back to the transmit end. The feedback includes the optimal sending rate derived from receive bandwidth and active flow quantity, enabling the transmit end to dynamically adjust its sending rate in real-time.

Inventive Principle:

Principle #23Feedback

2Reliability

If the transmit end decreases sending rate when network is congested, then congestion is controlled, but congestion control efficiency is low due to slow adjustment

Engineering Contradiction:

Improvecongestion controlVSAvoidcongestion control efficiency

Core Design Contradiction:

ReliabilityVSProductivity

Solution Approach 1:

The patent implements feedback by establishing a bidirectional communication mechanism where the receive end continuously monitors its receiving state, calculates optimal sending rates, and feeds this information back to the transmit end. The feedback includes the optimal sending rate derived from receive bandwidth and active flow quantity, enabling the transmit end to dynamically adjust its sending rate in real-time.

Inventive Principle:

Principle #23Feedback

Solution Approach 2:

The patent applies preliminary action by having the receive end proactively calculate and feed back the optimal sending rate to the transmit end before congestion occurs. The receive end determines the optimal rate based on its receiving capability and active flow quantity, then sends this information in advance via feedback messages, allowing the transmit end to adjust immediately without waiting for congestion to manifest.

Inventive Principle:

Principle #10Preliminary action

3Loss of information

If the receive end feeds back congestion information to the transmit end, then the transmit end can adjust sending rate, but the adjustment is gradual and causes delay

Engineering Contradiction:

Improvecongestion information feedbackVSAvoidadjustment delay

Core Design Contradiction:

Loss of informationVSLoss of time

Solution Approach 1:

The patent applies parameter changes by transforming the feedback mechanism from simple congestion state indicators (congested/not congested) to quantitative optimal sending rate values. The receive end calculates the optimal rate using the formula: optimal rate = receive bandwidth / (1 + active flow quantity), and feeds back this precise parameter, enabling the transmit end to make immediate numerical adjustments rather than gradual qualitative changes.

Inventive Principle:

Principle #35Parameter changes

Data Source

PatentEP3852323B1Congestion control method, and network apparatus

Publication Date: 2025.04.02 HUAWEI TECH CO LTD

Figure 1~2
Figure 3
Figure 4

AI summary

This application provides a congestion control method and a network device. The method includes: receiving, by a first network device, a first message sent by a second network device, where the first message carries an active flow quantity, and the active flow quantity is a quantity determined by the second network device based on a data flow to which data packets received from the first network device belong; determining, by the first network device based on the active flow quantity and rated receiving bandwidth of the second network device, packet sending control information used to send the data flow to the second network device, where the packet sending control information indicates that actual receiving bandwidth of the second network device reaches the rated receiving bandwidth when the first network device sends the data flow to the second network device based on the packet sending control information; and sending, by the first network device, the data flow to the second network device based on the packet sending control information. This application can better control congestion, thereby reducing network packet loss.