Network Policing via Path-Specific Greediness Benchmarking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current network policing methods fail to effectively manage congestion by not considering the specific characteristics of the data path, leading to inefficient policing and potential misidentification of misbehaving flows, which can result in under-investment in Quality of Service (QoS) and unfair bandwidth allocation.
Innovation Solution
A path-specific policing method that determines a flow's greediness relative to a reference rate adaptation policy and adjusts policing based on the expected throughput of compliant flows under similar path conditions, allowing for more accurate identification and sanctioning of misbehaving flows.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If current network policing methods are used, then policing can be implemented, but policing accuracy deteriorates due to not considering path-specific characteristics
Solution Approach 1:
The patent applies local quality by making the policing decision dependent on local path characteristics (congestion level, round-trip time) rather than using a uniform policing approach. Each flow is evaluated against the expected behavior for its specific path conditions, allowing accurate identification of misbehaving flows while maintaining fairness.
2Measurement precision
If path-specific policing is implemented, then policing accuracy improves, but false positives increase due to strict enforcement
Solution Approach 1:
The patent applies dynamics by making the policing threshold adaptive rather than fixed. The expected throughput is dynamically calculated based on current path conditions (congestion level m and round-trip time T), allowing the system to adjust its expectations according to actual network state and reduce false positives.
Solution Approach 2:
The patent implements feedback by continuously monitoring actual flow throughput and comparing it against the dynamically calculated expected throughput based on path conditions. This closed-loop approach allows the system to learn from actual behavior and adjust policing decisions accordingly.
3Productivity
If TCP rate control algorithm is used, then congestion management improves, but adaptability deteriorates for applications that cannot tolerate rapid rate variation
Solution Approach 1:
The patent applies segmentation by separating the rate control function from the policing function. The TCP rate control algorithm handles congestion management for compatible applications, while the policing mechanism independently evaluates compliance and enforces policies. This allows different applications to be treated differently based on their congestion response characteristics.
Data Source
AI summary
Flow in a data network is policed by determining a measure of greediness of a flow through a node, comparing the measure of greediness with a measure indicative of acceptable greediness dependent on the expected greediness of a compliant flow experiencing substantially similar path conditions, and designating the flow for possible sanction in the event that the greediness is not in accordance with the acceptable greediness. Such allows for the policing of a data network when one or more fields in a data packet that carry information relating to a characterization of an end-to-end path and/or a downstream path of the data packet are used in order to police the data network.