Delay-Based Network Congestion Control Using Queue Management
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current TCP Reno congestion control algorithms become performance bottlenecks in networks with large bandwidth-delay products, experiencing slow packet increase, drastic decrease, oscillations, and instability due to binary congestion signals and inaccurate packet loss estimation.
Innovation Solution
A delay-based congestion control scheme using scalable queue size management, one-way queueing delay measurement, and alpha tuning methods, including queue control, scalable utility functions, dynamic and randomized alpha tuning, and equation-based alpha tuning, to stabilize network dynamics and maintain high utilization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If TCP Reno uses linear increase by one packet per RTT, then packet traffic control is simple, but the increase speed is too slow for high bandwidth systems
Solution Approach 1:
The patent changes the control parameter from linear increase (additive increase) to multiplicative increase, where the congestion window grows exponentially rather than linearly. This allows the algorithm to adapt to high bandwidth systems by increasing packet transmission rates much faster, while maintaining simplicity through a straightforward multiplicative update rule based on ACK receipts.
2Speed
If TCP Reno uses multiplicative decrease per loss event, then congestion response is fast, but the decrease is too drastic causing severe oscillations
Solution Approach 1:
The patent implements dynamic adjustment of the congestion window using a multiplicative increase factor that is modulated by a damping term. Instead of a fixed multiplicative decrease, the system dynamically adjusts the window size based on current network conditions, using a control parameter that adapts the aggressiveness of the response. This creates a more stable dynamic system that responds quickly to congestion while avoiding severe oscillations through continuous adjustment rather than binary increase/decrease steps.
3Device complexity
If TCP Reno uses binary congestion signal (packet loss), then congestion detection is simple, but oscillation is unavoidable
Solution Approach 1:
The patent applies partial action by using a damping term that partially reduces the congestion window based on the measured packet loss rate, rather than a full binary decrease. The damping factor is proportional to the square root of the packet loss rate, creating a graduated response that is neither fully aggressive nor completely passive. This partial adjustment smooths out the binary nature of traditional congestion signals, reducing oscillations while maintaining simplicity in the underlying detection mechanism.
4Productivity
If TCP Reno maintains large average congestion windows, then throughput is high, but equilibrium loss probability must be extremely small which is not practical
Solution Approach 1:
The patent changes the relationship between congestion window size and loss probability by introducing a multiplicative increase factor and damping term. Instead of requiring extremely small loss probabilities to maintain large windows, the system uses a controlled growth rate that allows larger windows to be sustained at practical loss rates. The multiplicative update rule with damping creates a new equilibrium point where higher throughput is achieved at achievable loss probabilities rather than requiring near-perfect delivery rates.
5Quantity of substance
If network bandwidth-delay product continues to grow, then network capacity increases, but TCP Reno becomes a performance bottleneck
Solution Approach 1:
The patent implements a dynamic congestion control algorithm that adapts to growing bandwidth-delay products by using multiplicative increase and damping-based adjustment. As network capacity increases, the multiplicative nature of the control algorithm allows the congestion window to grow proportionally faster, preventing the bottleneck effect that plagues linear increase algorithms. The dynamic adjustment mechanism maintains efficiency by continuously adapting the transmission rate to match the available network capacity, regardless of how large the bandwidth-delay product becomes.
Data Source
AI summary
The invention provides a congestion control scheme that is a delay based scheme that includes a scalable queue size and one-way queueing delay measurement to reduce network congestion. Queue size is managed by queue control, a scalable utility function, dynamic alpha tuning, and/or randomized alpha tuning. One-way queueing delay is accomplished by measuring backward queueing delay management using various methods of estimating the receiver clock period. Embodiments include estimating the receiver clock period using single sample and multiple sample periods. The system includes a method for detecting route change.