Network Buffer Threshold Control for Packet Congestion

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current buffer management techniques, such as tail drop and Active Queue Management (AQM) schemas like RED and CoDel, fail to effectively prevent bufferbloat and global synchronization, leading to inefficient use of network capacity and unfair packet dropping.

Innovation Solution

A method that sets a threshold for packet queues in communication networks, providing congestion notifications only to the first packet exceeding the threshold and ignoring subsequent violations during a timeout interval, thereby reducing unnecessary data rate adjustments and avoiding global synchronization.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If large buffers are used to fill transmission capacities, then network capacity utilization is improved, but queuing delays and jitter increase which harms real time applications

Engineering Contradiction:
Improvenetwork capacity utilizationVSAvoidqueuing delays
Core Design Contradiction:
ProductivityVSLoss of time

Solution Approach 1:

The patent applies preliminary action by proactively dropping packets before the buffer becomes completely full. The buffer monitors its fill level and initiates packet dropping when reaching a threshold, preventing bufferbloat before it occurs. This early intervention maintains transmission capacity utilization while avoiding excessive queuing delays.

Inventive Principle:
Principle #10Preliminary action

2Reliability

If tail drop is used to drop packets when buffer is full, then buffer overflow is prevented, but global synchronization occurs and transmission capacity is underutilized

Engineering Contradiction:
Improvebuffer overflow preventionVSAvoidtransmission capacity utilization
Core Design Contradiction:
ReliabilityVSProductivity

Solution Approach 1:

The patent implements preliminary action by dropping packets proactively when the buffer reaches a threshold level, before complete overflow occurs. This prevents buffer overflow while avoiding the global synchronization problem of tail drop, as packets are dropped individually based on buffer state rather than all at once when full.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent uses feedback by continuously monitoring the buffer fill level and using this information to make dynamic dropping decisions. The buffer state feedback triggers selective packet dropping to maintain optimal utilization without causing global synchronization of senders.

Inventive Principle:
Principle #23Feedback

3Reliability

If RED is used to drop packets based on average queue size, then congestion notification is provided, but many parameters need to be set which makes deployment difficult

Engineering Contradiction:
Improvecongestion notificationVSAvoidparameter configuration
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent extracts the complex parameter configuration requirement from the congestion management mechanism. Instead of using RED's multiple parameters (thresholds, max/min drop probabilities, queue length calculations), the invention uses a simple buffer fill-level threshold that is automatically determined by buffer capacity, eliminating the need for manual parameter tuning while maintaining effective congestion notification.

Inventive Principle:
Principle #2Taking out (Extraction)

4Reliability

If buffers are kept empty most of the time to cope with short term overloads, then overload absorption is improved, but transmission capacity is underutilized

Engineering Contradiction:
Improveoverload absorptionVSAvoidtransmission capacity utilization
Core Design Contradiction:
ReliabilityVSProductivity

Solution Approach 1:

The patent applies preliminary action by maintaining buffers at optimal fill levels through proactive packet dropping. Instead of keeping buffers empty for overload protection, the buffer is actively managed at a threshold level that allows high utilization while preventing overload, dropping packets before the buffer becomes problematic.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP2884707B1Method for controlling buffering of packets in a communication network
Publication Date: 2016.04.27 ALCATEL LUCENT SA
  • EP2884707B1 patent drawingFigure 1
  • EP2884707B1 patent drawingFigure 2
  • EP2884707B1 patent drawingFigure 3

AI summary

The present invention concerns a method for controlling buffering of packets (1) in a communication network. The method comprises to store packets (1) of one or more packet flows in a queue (2) of a buffer (3), to set a threshold (4) on the queue (2), to determine whether the queue (2) exceeds the threshold (4). If the queue (2) exceeds the threshold (4), a congestion notification on a first packet (1 a) of the queue (2) causing the exceeding of the threshold (4) is provided and a timeout interval (5) is started. Until expiry of the timeout interval (5), any threshold violation is ignored by not providing a congestion notification on further packets (1b) of the queue (2) causing the exceeding of the threshold (4) during the timeout interval (5). The present invention concerns further a network element (8) for a communication network, the network element (8) is adapted to store packets (1) of one or more packet flows in a queue (2) of a buffer (3), to set a threshold (4) on the queue (2), to determine whether the queue (2) exceeds the threshold (4) and, if the queue (2) exceeds the threshold (4), to provide a congestion notification on a first packet (1 a) of the queue (2) causing the exceeding of the threshold (4), to start a timeout interval (5) and to ignore any threshold violation until expiry of the timeout interval (5) by not providing a congestion notification on further packets (1b) of the queue (2) causing the exceeding of the threshold (4) during the timeout interval (5). The present invention concerns further a computer program product adapted to perform the above method, when executed on a control unit (9).