TCP Congestion Control Using Forward Delay Components
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional TCP congestion control mechanisms are inadequate for delay-sensitive and bursty data, such as those in Storage Area Networks, leading to packet drops and inefficient bandwidth usage, as they fail to effectively manage window sizes and network buffer occupation.
Innovation Solution
The proposed solution adjusts the maximum send window using forward queuing delay and maximum bandwidth parameters, focusing on forward congestion delay to control the congestion window and implement rate shaping, thereby bounding network buffer occupation and varying the congestion window based on recently acknowledged bytes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional TCP congestion control mechanisms are used, then network adaptability and flow control are provided, but packet drops occur and bandwidth usage is inefficient for delay-sensitive and bursty data
Solution Approach 1:
The patent changes the parameters used for congestion control from conventional RTT-based metrics to forward delay component and maximum bandwidth parameters. The maximum send window is adjusted using the formula: max_send_window = min(max_send_window, (maximum_bandwidth × forward_delay) / (1 - α)), where α is a smoothing factor. This parameter change allows the system to adapt to delay-sensitive and bursty data characteristics while maintaining reliability and improving bandwidth efficiency.
2Productivity
If the send window size is increased to improve throughput, then bandwidth utilization improves, but network buffer occupation increases causing packet drops
Solution Approach 1:
The patent implements feedback control by continuously monitoring the forward delay component and adjusting the maximum send window accordingly. The system uses the formula: max_send_window = min(max_send_window, (maximum_bandwidth × forward_delay) / (1 - α)), where the forward delay provides feedback about current network congestion conditions. This feedback mechanism dynamically adjusts the window size to maintain high throughput while preventing buffer overflow and packet drops.
3Adaptability or versatility
If the congestion window is adjusted based on round trip time, then network adaptability is provided, but delay-sensitive data transmission is inadequate due to reverse path delay interference
Solution Approach 1:
The patent extracts the forward delay component from the round trip time measurement by using timestamp comparison: forward_delay = (receive_timestamp - send_timestamp). This extraction removes the reverse path delay interference from the congestion control calculations. The system then uses only the forward delay component to adjust the congestion window, providing accurate adaptation to forward path congestion conditions without being influenced by reverse path variations.
4Reliability
If rate shaping is implemented to bound network buffer occupation, then packet drop probability decreases, but traffic flow smoothness requires complex window adjustment mechanisms
Solution Approach 1:
The patent implements dynamic window adjustment using the formula: max_send_window = min(max_send_window, (maximum_bandwidth × forward_delay) / (1 - α)). The system dynamically adapts the send window size based on real-time forward delay measurements and maximum bandwidth estimates. The smoothing factor α provides gradual adaptation, and the rate shaping is achieved through continuous dynamic adjustment rather than complex discrete control mechanisms, bounding network buffer occupation while maintaining manageable complexity.
Data Source
AI summary
According to the present disclosure, methods and apparatus are provided to improve the Transmission Control Protocol (TCP) for data such as delay sensitive or bursty data. A maximum send window is adjusted using forward queuing delay and maximum bandwidth parameters. Reverse queuing delay and the number of packets drops are not factored into generation of the maximum send window. Network buffer occupation is bounded and a congestion window is effectively varied using rate shaping and gradual size increases based at least partially on the number acknowledged packets.