Network Congestion Management via Backwards Notification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current network traffic management systems are inefficient in handling congestion, leading to increased network delays, potential application failures, and system crashes due to the inability to effectively distribute congestion management information across the network.
Innovation Solution
A system that utilizes a congestion point queue, a monitor to sample its state, a consolidated parameter generator to reduce feedback message size, and a feedback message generator to send consolidated congestion information back to the source, implementing a backwards congestion notification (BCN) mechanism to shift congestion from the network core to edge devices where resources are more available.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If detailed congestion management information is sent from congestion points to reaction points, then congestion control precision is improved, but network bandwidth is consumed and congestion is worsened
Solution Approach 1:
The patent combines multiple congestion management parameters (queue depth, drop probability, congestion duration) into a single consolidated feedback message. This merging of information reduces the bandwidth required for congestion notifications while maintaining sufficient precision for effective congestion control at reaction points
Solution Approach 2:
The patent extracts only the essential congestion state information needed for control decisions and transmits it back to reaction points. By selecting and transmitting only critical parameters rather than all possible congestion data, the system achieves effective congestion control with minimal bandwidth consumption
2Loss of information
If congestion information is propagated throughout the network, then congestion awareness is improved, but network delay increases
Solution Approach 1:
The patent implements a feedback mechanism where congestion points send notifications back to reaction points about congestion conditions. This feedback loop provides timely congestion awareness to sources without requiring continuous propagation of congestion information throughout the entire network, thus reducing delay while maintaining awareness
Solution Approach 2:
The patent enables reaction points to preemptively adjust their transmission rates based on received congestion feedback before severe congestion occurs. This preliminary action allows sources to slow down in advance, preventing congestion from developing fully and reducing overall network delay
3Reliability
If packet dropping probability is increased at congestion points, then transient congestion is alleviated, but sustained congestion worsens due to loss of information
Solution Approach 1:
The patent uses feedback messages to communicate congestion state and drop probability information from congestion points to reaction points. This feedback ensures that even when packets are dropped, the congestion information is preserved and transmitted back to sources, allowing them to adjust their behavior appropriately for both transient and sustained congestion scenarios
Solution Approach 2:
The patent introduces feedback messages as an intermediary mechanism that carries congestion information between congestion points and reaction points. This intermediary ensures that congestion information is not lost even when data packets are dropped, maintaining the reliability of congestion control for both transient and sustained congestion
Data Source
AI summary
A method and system to manage network congestion are provided. In one example embodiment, the system comprises a congestion point queue, a monitor to sample a state of the congestion point queue, a consolidated parameter generator to generate a consolidated feedback parameter, and a feedback message generator to generate a feedback message, utilizing the consolidated feedback parameter. The congestion point queue may be configured to queue messages from a reaction point to a congestion point. The state of the congestion point queue may be reflected by an equilibrium queue level, a queue offset, and a rate of change of a size of the congestion point queue. The equilibrium queue level may represent a particular predetermined size of the congestion point queue. The queue offset may represent a deviation from the equilibrium queue level. The consolidated feedback parameter may be generated to reflect the queue offset and the rate of change of the size of the congestion point queue.


