AI-Aided Flow Identification for Queue-Less Traffic Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing network devices struggle with efficient traffic control in the absence of priority queues, particularly in transitioning between LAN and WAN networks, and require improved flow classification to distinguish between different flow types without significant implementation cost.
Innovation Solution
A network device employing AI-aided flow identification and drop-based queue-less traffic control, using machine learning to identify specific flows and classify them into predefined types, with override indications for certain flows, and applying drop-based traffic control to manage bandwidth without priority queues.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional flow classification methods are used to distinguish between different flow types, then implementation cost is controlled, but flow classification accuracy deteriorates making it difficult to distinguish between lower-priority flows (e.g., file transfer flows) and higher-priority flows (e.g., low latency flows and video streaming flows)
Solution Approach 1:
The patent introduces an AI-aided flow identification circuit as an intermediary component that assists the flow classification circuit. This AI circuit processes packet data and generates flow-type override indications, acting as a mediator between raw packet data and the classification decision, thereby improving classification accuracy without requiring complete redesign of the entire classification system
Solution Approach 2:
The flow classification function is segmented into two independent parts: (1) traditional flow classification circuit that handles general classification based on flow rate and pre-defined types, and (2) AI-aided flow identification circuit that handles specific difficult-to-classify flows. This segmentation allows each component to be optimized independently, improving overall accuracy while controlling implementation cost by only adding AI capability where needed
2Reliability
If priority queues are used for traffic control in WAN networks, then quality of service is improved, but device complexity increases and the solution becomes inapplicable for LAN networks where bandwidth is much larger
Solution Approach 1:
Instead of using the traditional approach of creating priority queues and scheduling them (which works for WAN but adds complexity), the patent inverts the approach by using drop-based queue-less traffic control. The system directly controls traffic by probabilistically dropping packets based on flow type, eliminating the need for complex queue management while maintaining QoS differentiation
Solution Approach 2:
The patent changes the traffic control parameter from queue scheduling decisions to packet drop probability. By adjusting the drop probability parameter differently for different flow types (higher drop for low-priority flows, lower drop for high-priority flows), the system achieves QoS control without requiring multiple priority queues, thereby reducing device complexity while maintaining service quality
Data Source
AI summary
A network device includes AI aided flow identification circuit and flow classification circuit. The AI aided flow identification circuit identifies specific flows of only a portion of pre-defined flow types through machine learning, and generates a flow-type override indication for each of the specific flows. The flow classification circuit classifies each flow into one pre-defined flow type. The AI aided flow identification circuit generates a flow-type override indication for a first flow. In response to the flow-type override indication for the first flow, the flow classification circuit classifies the first flow into a first flow type regardless of a flow rate of the first flow. The AI aided flow identification circuit does not generate a flow-type override indication for a second flow, and the flow classification circuit classifies the second flow into the first flow type or a second flow type according to a flow rate of the second flow.