Waveform Independent Protocol Congestion Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current TCP protocols face limitations in lossy networks, susceptibility to intelligent jamming, and path dependence, primarily due to reliance on packet loss and round trip time as congestion indicators, leading to poor performance and throughput reduction.
Innovation Solution
The Waveform Independent Protocol (WIP) uses capacity measurements at the source and destination to estimate congestion independently of packet loss and path stability, adjusting the congestion window based on the difference between send and receive rates, eliminating dependency on packet loss and RTT, and allowing for aggressive throughput without competing with loss-based TCP protocols.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If TCP protocols use packet loss as congestion indicator, then congestion detection capability is improved, but performance in lossy networks deteriorates
Solution Approach 1:
The patent segments the congestion detection function from packet loss detection by introducing a separate capacity estimation mechanism. The source node estimates network capacity independently using timing information from packet transmissions and acknowledgments, rather than relying on packet loss signals. This segmentation allows the protocol to detect congestion through capacity utilization metrics while remaining resilient to channel losses.
Solution Approach 2:
The patent introduces timing information (transmission timestamps and acknowledgment timestamps) as an intermediary measurement mechanism. Instead of directly using packet loss as the congestion indicator, the system uses timing-based capacity estimation as an intermediate step that indirectly reflects network congestion state without being affected by channel losses.
2Reliability
If TCP protocols rely on packet loss feedback, then congestion control is achieved, but susceptibility to intelligent jamming increases
Solution Approach 1:
The patent extracts the congestion detection function from the packet loss feedback mechanism. By removing dependency on packet loss as the congestion indicator and replacing it with timing-based capacity estimation, the system maintains congestion control effectiveness while eliminating the vulnerability to jamming attacks that exploit packet loss feedback.
Solution Approach 2:
The patent converts the harmful effect of packet loss (which can be manipulated by jammers) into a benign measurement mechanism. By using timing information from successfully transmitted packets and their acknowledgments, the system derives capacity estimates that are unaffected by malicious packet interruption, effectively converting the loss-based feedback into a resilient timing-based feedback.
3Measurement precision
If TCP protocols use RTT as congestion indicator, then path-based congestion detection is improved, but performance in opportunistic routing deteriorates
Solution Approach 1:
The patent introduces dynamic capacity estimation that adapts to changing network paths. Instead of relying on static RTT measurements that assume stable paths, the system continuously estimates capacity based on timing information from each transmission round, allowing it to adapt to path changes in opportunistic routing while maintaining congestion detection accuracy.
Solution Approach 2:
The patent changes the measurement parameter from RTT (which is path-dependent) to capacity utilization (which is flow-dependent). By measuring the difference between transmission rate and acknowledged reception rate, the system obtains a congestion indicator that remains valid regardless of path changes, enabling effective congestion control in opportunistic routing scenarios.
Data Source
AI summary
A method and apparatus for controlling congestion of information packets in a computer network comprising at least a first node and a second node is provided. At each iteration of the method a transmission rate for information packets sent from the first node to the second node and a reception rate for responses received from the second node to the first node are determined. The rates may be determined from timestamps inserted into the information packets. The length of a congestion window is increased when a time average of the transmission rate is no greater than a time average of the reception rate and decreased when a time average of the transmission rate is greater than a time average of the reception rate by at least a threshold amount. Transmission of information packets is adjusted dependent upon the congestion window.


