Cable Network Modelling via Bayesian Loss Distribution Updates
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Legacy inventory records in telecommunications networks often inaccurately label cable segments, leading to errors in determining broadband service speeds, especially with technologies like VDSL and G.fast, where noise sensitivity increases due to longer copper or aluminium cable lengths.
Innovation Solution
A method to model cable segment composition by determining loss measurements, setting initial loss distributions, estimating observed losses through Bayesian updates using Gibbs sampling, and mapping loss distributions onto cable composition probabilities, allowing for accurate estimation of cable material, gauge, and length.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If physical inspection of cable segments is performed to confirm material composition, then measurement precision is improved, but device complexity and time consumption increase
Solution Approach 1:
The patent replaces physical inspection with signal-based measurement. By measuring signal loss characteristics and using Bayesian inference to determine cable composition, the system eliminates the need for manual physical inspection while achieving accurate material identification.
Solution Approach 2:
The patent introduces signal loss measurements as an intermediary to indirectly determine cable composition. Instead of directly inspecting cables, the system uses signal propagation characteristics as a mediator to infer material properties through Bayesian updating.
2Ease of operation
If generic or inferred cable records are used, then ease of operation is improved, but measurement precision deteriorates
Solution Approach 1:
The patent implements feedback through Bayesian updating. The system continuously refines cable composition estimates by comparing signal loss measurements with predicted loss profiles, automatically correcting generic or inferred records without manual intervention.
Solution Approach 2:
The patent transforms static generic cable records into dynamic, updated composition probabilities. By changing the representation from fixed generic labels to probability distributions updated with measured signal loss data, the system maintains ease of operation while improving precision.
3Measurement precision
If cable segment composition is accurately determined, then broadband service speed estimation is improved, but loss of time increases due to measurement and modelling requirements
Solution Approach 1:
The patent performs preliminary actions by establishing prior probability distributions for cable composition based on network knowledge before actual measurements are taken. This allows the system to quickly update estimates using minimal measurement data rather than requiring extensive measurement time.
Solution Approach 2:
The patent uses partial measurements (signal loss at specific frequencies) combined with Bayesian inference to achieve accurate composition determination without requiring complete physical inspection. The partial measurements provide sufficient information to update probability distributions effectively.
Data Source
AI summary
The invention generates a model of the composition of the cable segments in a network. Different cable compositions have different loss distributions. Each cable segment is given a starting cable composition (based on cable records if available), and thus can be represented as a loss distribution. The loss for each circuit can be measured (by measuring Hlog), and thus can also be represented as a loss distribution. Updates are made to the loss distribution for each segment so that the loss distributions that make up each circuit is consistent with that of the (measured) loss distribution for that circuit. These updates are preferably performed as Bayesian updates of each cable segment (loss distribution) using Gibbs sampling (i.e. the other cable segment loss probabilities are fixed whilst the probability for the segment under consideration is updated).