Vectored-DSL Filter Training Exit via Convergence Thresholds
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In Vectored-Digital Subscriber Line (DSL) technology, high frequency crosstalk leads to unstable performance and low line activation rates, with existing methods failing to efficiently exit the training procedure without interfering with adjacent subscribers or causing call drops.
Innovation Solution
A method and system that rapidly exit the training procedure by comparing feedback error values with a preset threshold, using pairwise orthogonal sequences to calculate and acquire feedback error values, and exiting the training when convergence conditions are met, thereby reducing interference and shortening training time.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a fixed and high training time is used for the precoder or canceller, then the crosstalk cancellation performance is improved, but the interference to adjacent subscribers increases and call drops occur
Solution Approach 1:
The patent applies dynamics by transitioning from a fixed training time to a dynamic training time that adapts to convergence conditions. The training procedure continuously monitors feedback error values and adjusts the training duration based on whether convergence criteria are met, allowing the system to exit training early when performance is sufficient, thereby reducing interference to adjacent subscribers while maintaining adequate crosstalk cancellation performance.
Solution Approach 2:
The patent implements feedback mechanisms by continuously monitoring feedback error values during the training procedure. The DSLAM receives feedback from subscribers about the quality of received signals and uses this information to determine whether convergence conditions are met. This feedback-driven approach enables the system to make informed decisions about when to exit training, balancing performance requirements with interference reduction.
2Manufacturing precision
If a fixed and high training time is used for the precoder or canceller, then the filtering performance is improved, but the training time is extended and interference to other subscribers increases
Solution Approach 1:
The system dynamically adjusts training duration based on real-time convergence assessment. Instead of using a predetermined fixed training time, the system continuously evaluates feedback error values and exits training as soon as convergence conditions are satisfied. This dynamic approach reduces unnecessary training time while maintaining adequate filtering performance.
Solution Approach 2:
The training procedure becomes self-regulating by automatically monitoring its own performance through feedback error values and making autonomous decisions about when to terminate training. The system serves itself by detecting convergence conditions and initiating exit from training without requiring external intervention or fixed timing parameters.
3Measurement precision
If the training procedure is extended to improve convergence, then the filtering accuracy is improved, but the probability of call drops of other subscribers increases
Solution Approach 1:
The system uses feedback from subscribers during training to assess convergence status in real-time. By monitoring feedback error values and comparing them against convergence thresholds, the system can determine when adequate filtering accuracy has been achieved and exit training promptly, preventing prolonged interference that would cause call drops in other subscribers.
Solution Approach 2:
The patent applies the skipping principle by allowing the training procedure to be terminated early once convergence conditions are met, rather than completing a full fixed-duration training cycle. This approach rushes through the unnecessary portion of training that would not contribute to further performance improvement, thereby reducing the window of interference exposure and minimizing call drop probability.
Data Source
Figure 1
Figure 2
Figure 3~4
AI summary
A method, a system, and a device for rapidly exiting training are provided. The method includes: in a training procedure of a Vectored-Digital Subscriber Line (DSL) filter, comparing a feedback error value of the filter and/or a swing range of a filter coefficient with a preset threshold, and determining whether the filter meets a convergence condition according to a comparison result; and exiting the training procedure of the filter when the filter meets the convergence condition. The method, system, and device are applicable to a training procedure of a Vectored-DSL precoder or canceller, so as to rapidly exit the training procedure.