Exponential Packet Select Probability for Queue Congestion Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The conventional RED algorithm for active queue management is too aggressive when the queue length exceeds the minimum threshold by a small amount and too lenient when close to the maximum threshold, leading to inefficient packet discard/select rates.
Innovation Solution
The proposed method calculates a packet select probability using an exponential function based on queue congestion, providing a more lenient rate at lower congestion values and a more aggressive rate at higher values, adjusting the aggressiveness factor to optimize performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If the conventional RED algorithm uses a linear probability increase from minimum threshold to maximum threshold, then the packet discard rate is aggressive when queue length exceeds minimum threshold by a small amount, but too lenient when close to maximum threshold
Solution Approach 1:
The patent changes the functional form of the probability parameter from linear to exponential. Specifically, it uses Pselect = e^(-α·Qcon) where Qcon is the queue congestion value and α is an aggressiveness factor. This exponential transformation allows the system to achieve both leniency at low congestion and aggressiveness at high congestion without requiring complex multi-threshold logic, thus improving packet discard rate efficiency while maintaining calculation simplicity.
2Reliability
If the RED algorithm drops packets early to prevent buffer overrun, then queue utilization is reduced, but packet loss increases
Solution Approach 1:
The patent implements a feedback mechanism where the packet select probability is dynamically adjusted based on real-time queue congestion measurement. The system continuously monitors queue length, calculates congestion value Qcon, and adjusts Pselect accordingly using the exponential function. This feedback loop allows the system to apply gentle pressure early (low probability) and aggressive pressure later (high probability), optimizing the balance between preventing buffer overrun and minimizing unnecessary packet loss.
3Stability of the object's composition
If the RED algorithm uses randomization in packet dropping, then global synchronization between traffic flows is avoided, but unfairness to bursty traffic flows occurs
Solution Approach 1:
The patent introduces dynamic adjustment of the aggressiveness factor α based on traffic flow characteristics and queue conditions. By making α variable rather than fixed, the system can adapt its packet selection behavior to different traffic patterns. This dynamic approach allows the system to be more lenient to bursty flows during certain conditions while maintaining the randomization benefit of avoiding global synchronization, thus balancing both requirements.
Data Source
AI summary
A method and apparatus for deriving a packet select probability value for a data packet. The method comprises determining a queue length value for a target buffer of the data packet, calculating a queue congestion value based at least partly on the queue length value and a packet select queue length range, and calculating the packet select probability value for the data packet based at least partly on an exponential function e−x, where x is computed based at least partly on the queue congestion value.