Nonlinear Term Selection for Volterra Model Complexity Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The complexity of Volterra series expansion models in communication systems increases exponentially with memory length and order, leading to over-training and reduced precision due to the inclusion of unnecessary nonlinear terms.
Innovation Solution
A nonlinear term selection apparatus and method that uses linear coefficients to selectively discard terms of lesser contribution and retain those of greater significance, simplifying the model and reducing complexity by normalizing and comparing coefficients against a threshold value.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If all Volterra series expansion terms are included in the nonlinear model, then the model completeness is improved, but the model complexity increases exponentially
Solution Approach 1:
The patent extracts and removes unnecessary nonlinear terms from the Volterra series expansion based on coefficient magnitude thresholds. By calculating linear coefficients first and using them to identify significant nonlinear terms, the method extracts only the essential terms needed for accurate modeling, thereby reducing model complexity while preserving completeness.
Solution Approach 2:
The patent changes the parameter selection criterion from including all terms to including only terms whose coefficients exceed a threshold value. This parameter-based filtering approach transforms the model construction process, allowing the system to adaptively select terms based on their actual contribution to the model accuracy.
2Reliability
If more nonlinear terms are included in the model, then the model coverage is improved, but the training precision deteriorates due to over-training
Solution Approach 1:
The patent removes redundant nonlinear terms that contribute minimally to model coverage. By calculating coefficients and comparing them against thresholds, the method extracts only the significant terms, preventing over-training while maintaining adequate model coverage.
Solution Approach 2:
Instead of including all possible nonlinear terms (excessive action), the patent selectively includes only those terms whose coefficients exceed a threshold (partial action). This partial inclusion strategy prevents over-training while achieving sufficient model coverage for practical applications.
3Device complexity
If the number of nonlinear terms is reduced, then the model complexity is lowered, but the model precision may deteriorate
Solution Approach 1:
The patent uses coefficient magnitude as a parameter to determine term inclusion. By setting an appropriate threshold, the method ensures that only terms with sufficient contribution to precision are retained, thus reducing complexity without sacrificing model precision.
Solution Approach 2:
The patent implements a feedback mechanism where linear coefficients are calculated first, then used to guide the selection of nonlinear terms. This feedback loop ensures that term reduction does not compromise precision, as the selection is based on actual measured coefficient magnitudes rather than arbitrary criteria.
Data Source
AI summary
The embodiments of the present invention provide a nonlinear term selection apparatus and method, an identification system and a compensation system. The selection apparatus comprises: a linear coefficient calculator configured to measure linear properties of a nonlinear system by using measurement data, so as to obtain a plurality of linear coefficients; and a nonlinear term selector configured to select nonlinear model expanded terms of the nonlinear system by using the plurality of linear coefficients, so as to obtain nonlinear terms of the nonlinear system. With the embodiments of the present invention, the nonlinear model may be simplified, and the complexity of the nonlinear model may be lowered.


