Network Topology Discovery Using Confidence Value Aggregation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing network topology discovery methods face accuracy issues when using single types of network characteristic data, especially when the network does not support the data type or when data is incomplete or inconsistent, and lack comprehensive analysis when using multiple data types.
Innovation Solution
A method that collects network characteristic data from all network elements, uses multiple topology discovery algorithms to obtain link subsets, combines same links, calculates confidence values using an uncertainty reasoning algorithm, and retains the link with the highest confidence value for improved accuracy, ultimately generating a comprehensive network topology.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If network topology discovery is performed based on a single type of network characteristic data, then the discovery process is simple, but accuracy of network topology discovery is low
Solution Approach 1:
The patent merges multiple topology discovery algorithms (e.g., IP address matching, port alias matching, CDP) into a unified framework. Each algorithm processes different types of network characteristic data, and their results are combined through confidence value aggregation to produce a more accurate network topology than any single algorithm could achieve alone.
Solution Approach 2:
The patent creates a universal topology discovery system that can handle multiple types of network characteristic data and multiple discovery algorithms through a single integrated platform. The system assigns confidence values to results from different algorithms and combines them, making the system multi-functional and adaptable to various network environments.
2Reliability
If multiple types of network characteristic data are used for topology discovery, then comprehensive analysis can be performed, but the system complexity increases
Solution Approach 1:
The patent introduces confidence values as a parameter to quantify the reliability of topology discovery results from different algorithms. By assigning and aggregating confidence values, the system can objectively evaluate and combine results from multiple data types, improving reliability while managing complexity through standardized parameter evaluation.
3Measurement precision
If data is not completely collected or data is inconsistent in the network, then topology discovery can still proceed, but uncertainty exists in link relationship and accuracy is low
Solution Approach 1:
The patent prepares for potential data incompleteness or inconsistency by assigning confidence values to each discovered link based on the quality and consistency of supporting evidence from multiple algorithms. This cushioning approach allows the system to handle uncertain or incomplete data gracefully, maintaining accuracy by weighting results according to their reliability rather than failing outright.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Embodiments of the present invention provide a network topology discovery method and device, which can perform a comprehensive analysis on results obtained after network topology discovery is performed by using multiple types of network characteristic data, and improve accuracy of network topology discovery. A specific solution is: first, collecting network characteristic data of a network element in a to-be-analyzed network; then, obtaining at least two corresponding link subsets respectively by using at least two types of topology discovery algorithms and according to the network characteristic data, gathering all links in the at least two link subsets into one set to obtain a first link set, combining same links in the first link set, and for multiple links having only one same port, retaining only a link having a largest confidence value in the multiple links and deleting a remaining link, to obtain a second link set; and last, obtaining a network topology of the to-be-analyzed network according to the second link set. The embodiments of the present invention are used for network topology discovery.