Waveform Equalization Using Truncated Volterra Kernels
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional digital signal processing methods for waveform equalization, such as those using the Volterra series, face high computational complexity and a large number of taps, making real-time processing challenging, especially when dealing with nonlinearity in optical fiber communications.
Innovation Solution
A waveform equalization device that employs a second-order Volterra series with a reduced kernel number (h) to simplify the equalization processing, using the formula y(m) = w1T·x1,m + ∑n=0h-1 w2,nT·(x2,m·x2,m-n), where h is the total number of kernels considered, significantly reducing computational complexity and the number of taps.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional Volterra series equalization is used, then waveform equalization performance is improved, but computational complexity increases
Solution Approach 1:
The patent extracts only the essential nonlinear components from the complete Volterra series expansion. By identifying and retaining only the dominant second-order nonlinear terms that contribute most to waveform distortion correction, while eliminating higher-order and redundant terms, the method achieves effective equalization with significantly reduced computational complexity.
Solution Approach 2:
The patent applies partial action by implementing a truncated Volterra series that includes only necessary terms (linear term and selected second-order nonlinear terms) rather than the complete expansion. This selective inclusion of partial terms provides sufficient equalization performance for practical systems while avoiding the excessive computational burden of calculating all possible higher-order terms.
2Measurement precision
If memory length L2 is increased, then equalization accuracy is improved, but the number of taps and computational complexity increase
Solution Approach 1:
The patent extracts and retains only the most significant tap coefficients from the complete Volterra expansion. By analyzing the contribution of each tap to the overall equalization performance, the method selectively keeps taps with substantial weights while discarding those with minimal impact, thereby maintaining accuracy with fewer taps.
Solution Approach 2:
The patent changes the parameter of memory length from a fixed large value to an optimized smaller value. Through analysis of the Volterra series convergence properties and practical system requirements, the method determines an optimal memory length that provides sufficient equalization accuracy while keeping the number of taps manageable for real-time processing.
Data Source
AI summary
An inference processing apparatus includes an input data storage unit that stores pieces X of input data, a learned NN storage unit that stores a piece W of weight data of a neural network, a batch processing control unit that sets a batch size on the basis of information on the pieces X of input data, a memory control unit that reads out, from the input data storage unit, the pieces X of input data corresponding to the set batch size, and an inference operation unit that batch-processes operation in the neural network using, as input, the pieces X of input data corresponding to the batch size and the piece W of weight data and infers a feature of the pieces X of input data.