ANN Equalizer with Bidirectional Inference for High-Speed Channels
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Data signals transmitted through channels experience distortion due to factors like skin effect and dielectric loss, leading to intersymbol interference and noise, especially at high speeds, which existing equalizers struggle to effectively compensate for.
Innovation Solution
An equalizer incorporating an artificial neural network (ANN) structure with a buffer circuit, forward pass circuit, and backward pass circuit to generate equalized output data by performing inference in both feedforward and feedback manners, compensating for signal distortion without recursive circuits.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If high-speed data transmission is performed through channels, then data transmission speed increases, but signal distortion and intersymbol interference worsen due to skin effect and dielectric loss
Solution Approach 1:
The equalizer is divided into multiple processing stages including a buffer circuit for storing samples, a forward pass circuit for feedforward processing, and a backward pass circuit for feedback processing. Each stage handles specific aspects of signal distortion compensation, allowing the system to maintain high transmission speeds while effectively addressing signal quality degradation through distributed processing.
Solution Approach 2:
The buffer circuit stores sequential samples before they are processed by the forward and backward pass circuits. This preliminary storage allows the system to access both past and future samples for inference, enabling proactive compensation for signal distortion before it fully degrades the output signal quality.
2Reliability
If conventional equalizers are used to compensate for signal distortion, then signal quality improves, but they fail to effectively handle high-speed transmission distortion
Solution Approach 1:
The backward pass circuit performs inference in a feedback manner, utilizing stored samples to generate corrections that are applied to compensate for signal distortion. This feedback mechanism allows the equalizer to adapt to high-speed transmission characteristics and effectively reduce intersymbol interference, maintaining signal quality at speeds where conventional equalizers fail.
Solution Approach 2:
The equalizer employs dynamic inference processes in both forward and backward passes, allowing it to adapt its processing based on the specific characteristics of the input signal. This dynamic approach enables the system to maintain high adaptability across different transmission speeds and channel conditions, overcoming the limitations of static conventional equalizers.
3Measurement precision
If ANN-based inference is performed in both forward and backward directions, then signal distortion compensation accuracy improves, but computational complexity increases
Solution Approach 1:
The forward pass circuit and backward pass circuit are merged into a unified equalizer architecture that shares the buffer circuit and processing elements. This merging allows the system to perform both feedforward and feedback inference using the same hardware resources, achieving high distortion compensation accuracy while reducing overall computational complexity compared to separate independent processing systems.
Data Source
AI summary
An equalizer, an operating method of the equalizer, and a receiver including the equalizer are provided. An equalizer including an artificial neural network (ANN) structure includes a buffer circuit configured to store samples sequentially input thereto and provide forward data and backward data, a forward pass circuit configured to generate a forward output by performing inference from the forward data in a feedforward manner according to an order in which the samples are input, a backward pass circuit configured to generate a backward output by performing inference from the backward data in a feedback manner in a reverse order to the order in which the samples are input, and a merging circuit configured to generate equalized output data, based on the forward output and the backward output.