Switching Device Data Packet Marking for Queue Congestion Control

Solution Overview

Problem

In data transmission systems, high delays occur due to excessive congestion notification (ECN) marking, which reduces the transmission efficiency of all data flows and causes unnecessary delays, especially when the length of the transmit queue exceeds a threshold.

Innovation Solution

A data packet marking method that dynamically determines a marking probability based on target and auxiliary parameters of a data flow, such as its level and transmission rate, to selectively perform congestion marking, thereby reducing the likelihood of unnecessary ECN marking on data flows with small amounts of data and minimizing delays.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If ECN marking is performed on data packets in each data flow when transmit queue length exceeds threshold, then queue congestion is addressed, but transmission efficiency of all data flows is reduced and delay increases

Engineering Contradiction:

Improvequeue congestion controlVSAvoidtransmission efficiency

Core Design Contradiction:

ReliabilityVSProductivity

Solution Approach 1:

The patent applies local quality by differentiating treatment based on data flow characteristics. Instead of uniformly marking all data flows when queue congestion occurs, the system calculates a marking probability for each flow based on its packet size distribution and only marks flows exceeding a size threshold. This selective local marking maintains congestion control effectiveness while preserving transmission efficiency for smaller flows.

Inventive Principle:

Principle #3Local quality

Solution Approach 2:

The patent changes the parameter from binary marking (mark all or mark none) to probabilistic marking based on flow-specific parameters. The marking probability is dynamically adjusted according to each data flow's packet size distribution characteristics, allowing fine-grained control that balances congestion management with transmission efficiency.

Inventive Principle:

Principle #35Parameter changes

2Reliability

If ECN marking is performed on all data flows when transmit queue length exceeds threshold, then congestion is notified, but unnecessary delays occur for data flows with small amounts of data

Engineering Contradiction:

Improvecongestion notificationVSAvoiddata flow delay

Core Design Contradiction:

ReliabilityVSLoss of time

Solution Approach 1:

The patent applies local quality by differentiating treatment based on data flow characteristics. Instead of uniformly marking all data flows when queue congestion occurs, the system calculates a marking probability for each flow based on its packet size distribution and only marks flows exceeding a size threshold. This selective local marking maintains congestion control effectiveness while preserving transmission efficiency for smaller flows.

Inventive Principle:

Principle #3Local quality

Solution Approach 2:

The patent applies partial action by marking only a subset of data flows (those exceeding the size threshold) rather than all flows. This partial marking approach provides sufficient congestion notification for the most impactful flows while avoiding unnecessary delays for smaller flows that contribute less to overall congestion.

Inventive Principle:

Principle #16Partial or excessive action

3Device complexity

If marking probability is determined based on target parameter only, then calculation is simple, but marking accuracy for different data flow levels is insufficient

Engineering Contradiction:

Improvemarking probability calculationVSAvoidmarking probability accuracy

Core Design Contradiction:

Device complexityVSMeasurement precision

Solution Approach 1:

The patent applies local quality by differentiating treatment based on data flow characteristics. Instead of uniformly marking all data flows when queue congestion occurs, the system calculates a marking probability for each flow based on its packet size distribution and only marks flows exceeding a size threshold. This selective local marking maintains congestion control effectiveness while preserving transmission efficiency for smaller flows.

Inventive Principle:

Principle #3Local quality

Solution Approach 2:

The patent adds another dimension to the marking probability calculation by incorporating data flow level information. The system determines marking probability based on both target parameter (packet size distribution) and auxiliary parameter (flow level), creating a two-dimensional assessment that improves marking accuracy while maintaining reasonable computational complexity.

Inventive Principle:

Principle #17Another dimension (Dimensionality change)

Data Source

PatentUS11991085B2Data packet marking method and device, and data transmission system

Publication Date: 2024.05.21 HUAWEI TECH CO LTD

AI summary

This application discloses: collecting statistics about a target parameter of a first data flow, where a target queue of a switching device is used to buffer a data packet in at least one data flow, the first data flow is one of the at least one data flow, and the target parameter is used to reflect an amount of data in the first data flow; when a length of the target queue meets a first length condition, determining, based on at least one of the target parameter and an auxiliary parameter of the first data flow, a marking probability corresponding to the first data flow; and performing congestion marking on a data packet in the first data flow based on the marking probability corresponding to the first data flow.