Network Fault Detection via Principal Component Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for monitoring data communications performance in broadband IP networks, particularly over legacy copper local access networks, face challenges in detecting fault conditions in a timely and cost-effective manner, often requiring expensive equipment with limited coverage and response time.
Innovation Solution
A method and apparatus using principal component analysis to model data communications performance by retrieving parameter data from network termination devices, projecting it into a coordinate system, and generating alerts for significant deviations, allowing for remote monitoring and fault detection across a network, including copper lines, with a focus on customer service experience.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If expensive monitoring equipment is used, then measurement precision is improved, but device complexity and cost increase
Solution Approach 1:
The patent creates a virtual model (copy) of the copper network's electrical characteristics through principal component analysis. Instead of using expensive physical monitoring equipment at each location, the system generates a mathematical representation that replicates the network's behavior, allowing fault detection through software-based analysis of this virtual model rather than through complex hardware installations.
Solution Approach 2:
The patent replaces physical monitoring hardware with a software-based mathematical model. The principal component analysis creates a virtual representation of the network that can be analyzed computationally, substituting mechanical/electrical monitoring devices with algorithmic processing that achieves similar or superior detection capabilities without the associated hardware complexity.
2Reliability
If comprehensive network monitoring is implemented, then reliability is improved, but loss of time for data collection and analysis increases
Solution Approach 1:
The patent performs preliminary action by creating the principal component model during an initial training phase using historical data. This pre-established model captures the normal electrical characteristics of the copper network, so when faults occur, the system can immediately compare current measurements against this pre-computed reference, eliminating the need for real-time complex analysis and enabling rapid fault detection.
Solution Approach 2:
The patent uses principal component analysis to extract only the most significant features from the network data, focusing computational resources on the key parameters that indicate faults. Rather than analyzing all possible network parameters equally, the system identifies and monitors only the critical dimensions of network behavior, reducing computational overhead while maintaining detection effectiveness.
3Area of stationary object
If remote monitoring of multiple users is performed, then coverage area is improved, but device complexity increases
Solution Approach 1:
The patent creates a universal monitoring system based on the electrical characteristics of copper networks that can be applied across multiple users and locations. The principal component model captures the fundamental behavior of copper line communications, which is consistent across different installations. This universal approach allows the same monitoring methodology to be deployed network-wide without requiring user-specific customization, achieving broad coverage through a standardized system.
Solution Approach 2:
The patent merges data from multiple network users into a unified principal component model. By aggregating electrical characteristic data across numerous users and applying dimensionality reduction, the system creates a consolidated view of network health that spans the entire copper infrastructure. This merging approach enables comprehensive monitoring coverage while simplifying the system architecture compared to implementing separate monitoring solutions for each user.
Data Source
AI summary
A method and apparatus are provided for identifying and locating fault conditions in a network. Network performance data are derived from a network termination device perspective which enables fault detection in any part or parts of a link between the device and broadband remote access server (BRAS) including copper lines. For example, during a training phase, parameter data generated by a device is retrieved for a plurality of network termination devices in a predetermined portion of the network. The parameter data comprising values for a predetermined set of parameters indicative of data communications performance over that portion of the network. The data is analyzed and one or more models are generated. In a monitoring phase, the same parameters are retrieved from a network termination device of a selected user served by the same portion of the network to expose any significantly divergent behavior, potentially indicative of a fault condition.