Network Node Packet Marking for Mixed Scalable Congestion Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing network communication nodes experience traffic congestion due to varying queuing delays and inefficiencies in managing scalable and unscalable congestion control, leading to latency and resource imbalances.
Innovation Solution
A method that classifies packets as scalable or unscalable based on identifiers, applies congestion marks using a probability-based marking ratio, and adjusts marking densities to balance traffic in a shared queue, ensuring fair resource distribution and low latency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional multiplicative reduction of packet transmission rate is applied, then congestion control is achieved, but queuing delay and utilization vary substantially when traffic flow rate changes
Solution Approach 1:
The patent implements dynamic congestion control by transitioning from static multiplicative reduction to dynamic scalable congestion control where the packet transmission rate is reduced proportionally to the amount of received marked packets. This allows the system to adapt its response based on current congestion levels and traffic flow rates, maintaining low queuing delay while effectively controlling congestion.
Solution Approach 2:
The patent changes the control parameter from multiplicative reduction factor to proportional reduction based on marked packet count. This parameter change enables the system to scale its congestion control response according to traffic flow rate, ensuring that queuing delay remains low even as traffic scales while maintaining effective congestion control.
2Loss of time
If scalable congestion control is implemented, then low latency communication is achieved with limited queuing delay, but substantial adaptations to senders, nodes, and receivers are required
Solution Approach 1:
The patent segments the network traffic into scalable packets and other packets based on identifiers, allowing different congestion control mechanisms to be applied to different packet types. This segmentation enables gradual transition where nodes can handle both scalable and traditional unscalable congestion control packets simultaneously, reducing the burden of substantial system-wide adaptations.
Solution Approach 2:
The patent creates a universal congestion control mechanism that can handle both scalable and unscalable congestion control packets in the same network node. The node maintains a marking probability based on marking ratio and applies appropriate marking strategies to different packet types, making the system multi-functional and compatible with both congestion control paradigms without requiring complete system redesign.
3Productivity
If a single queue is shared between scalable and unscalable packets, then resource utilization improves, but fairness in resource distribution becomes challenging
Solution Approach 1:
The patent applies different marking strategies to different packet types within the same queue. Scalable packets receive marking based on one strategy while unscalable packets receive marking based on another strategy. This local differentiation ensures that each packet type receives appropriate congestion control treatment, maintaining fairness while allowing both types to share the same queue resources efficiently.
Solution Approach 2:
The patent introduces a marking probability mechanism as an intermediary that mediates between scalable and unscalable congestion control requirements. The marking probability, based on the marking ratio of congestion-causing packets, acts as a mediator to fairly distribute congestion control signals to both packet types, ensuring fair resource distribution while maintaining high queue utilization.
Data Source
Figure 1
Figure 2
Figure 3A
AI summary
Example embodiments describe a method for managing data traffic congestion in a network communication node. The method comprising maintaining (202) a marking probability (223) based on a marking ratio indicative for a ratio between a change in packets responsible for congestion (221) in a network queue (122) in the network communication node, and a change in a total number of packets processed (222) by the network queue. The method further comprising classifying (201) packets processed by the network queue as scalable packets or other packets based on an identifier (212) included in the packets (210); wherein the scalable packets support scalable congestion control. The method further comprising marking (203) the scalable packets responsible for congestion in the network queue with a congestion mark (213) to signal the data traffic congestion; and marking the other packets with the congestion mark based on the marking probability if supporting unscalable congestion control.