Network Congestion Detection Using Representative Latency Metrics
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current network management systems lack effective methods to accurately predict and manage network latency and congestion levels, leading to potential bottlenecks and performance issues across various network devices and endpoints.
Innovation Solution
A system comprising a device manager with a processor that computes representative latency values and determines congestion levels by analyzing latency values for packets passing through network devices, using these metrics to generate indications for output and potentially apply recommendations for network optimization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If network management systems use traditional monitoring methods, then system complexity is reduced, but measurement precision of latency and congestion levels deteriorates
Solution Approach 1:
The patent introduces representative latency values as an intermediary metric that mediates between individual packet latency measurements and overall congestion assessment. This intermediary layer enables precise congestion prediction without requiring complex analysis of every single packet, thus improving measurement precision while managing system complexity.
Solution Approach 2:
The system performs preliminary computation of representative latency values from historical packet data before actual congestion occurs. This preliminary action establishes baseline metrics that enable early prediction of congestion conditions, improving measurement precision by having pre-computed reference values ready for comparison.
2Measurement precision
If network administrators manually monitor and manage network devices, then measurement precision of congestion levels is improved, but loss of time increases
Solution Approach 1:
The system enables self-service by automatically computing representative latency values and generating congestion predictions without requiring manual administrator intervention. The automated comparison of current latency against representative values and threshold-based congestion detection eliminates time-consuming manual monitoring while maintaining precise measurement capabilities.
Solution Approach 2:
The patent implements feedback mechanisms where congestion predictions are continuously generated and can trigger automated responses or alerts. This feedback loop enables rapid detection and response to congestion conditions, significantly reducing the time loss associated with manual monitoring by providing real-time automated assessments.
3Measurement precision
If network systems implement real-time latency monitoring for all packets, then measurement precision improves, but use of energy increases
Solution Approach 1:
The patent extracts only the essential information needed for congestion prediction by computing representative latency values from a subset or sampled packet data rather than analyzing every single packet in real-time. This extraction approach maintains sufficient measurement precision for congestion detection while dramatically reducing the energy consumption associated with processing all network traffic.
Solution Approach 2:
The system applies partial action by monitoring representative samples of network traffic rather than all packets. The representative latency values are computed from selected packet data that provides sufficient statistical significance for congestion prediction, achieving adequate measurement precision with reduced energy expenditure compared to exhaustive real-time monitoring of every packet.
4Device complexity
If network management systems use simplified latency metrics, then device complexity is reduced, but measurement precision of congestion prediction deteriorates
Solution Approach 1:
The patent transforms raw packet latency data into a changed parameter form - representative latency values - that are specifically tailored for congestion prediction. This parameter transformation maintains measurement precision by preserving the essential congestion-indicating characteristics of the original data while simplifying the form for easier comparison and analysis against thresholds.
Solution Approach 2:
The system applies local quality by making different parts of the data processing pipeline serve different purposes: raw packet latency data is collected with high fidelity, then transformed into representative values with specific statistical properties optimized for congestion detection. This localized optimization of data representation maintains precision where needed while reducing complexity in the analysis stage.
Data Source
AI summary
In one embodiment, a system comprising memory and processor(s), the processor(s) at least adapted to compute representative latency value(s) for a network device, based on a plurality of latency values obtained for a plurality of packets which passed through the network device over a period of time, the plurality of latency values indicative of latency between ingress to and egress from the network device, obtain latency value(s) for packet(s) which passed through the network device after the period of time, the latency value(s) indicative of latency between ingress to and egress from the network device, determine a determination of congestion level(s) for the network device, including analyzing at least one of the latency value(s) in relation to at least one of the representative latency value(s), and generate an indication in accordance with the determination, the indication to be outputted on at least one output device.