SERDES Training via Inverse Channel Impulse Response Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional SERDES devices face challenges in reliable training due to the need for equalization, which requires reliable capture of training bits, creating a circular dependency and making it difficult to properly train the device.
Innovation Solution
The method involves transmitting a Common Electrical Interface (CEI) training frame to determine the effective aggregate channel impulse response, estimating its inverse, and using this inverse to compensate for distortions introduced during signal transmission, allowing for effective equalization either at the transmitter or receiver side.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If equalization is performed to reduce transmission distortions, then signal quality is improved, but reliable capture of training bits becomes difficult creating a circular dependency
Solution Approach 1:
The patent applies preliminary action by pre-processing the training signal through known test patterns and predetermined equalizer settings before transmission. The receiver uses these known patterns to establish initial channel estimates and equalizer coefficients, breaking the circular dependency by having equalization partially performed in advance based on known information.
Solution Approach 2:
The patent implements feedback by using the received training signal to estimate channel characteristics and feed this information back to adjust equalizer settings. The system continuously monitors signal quality metrics from the training sequence and uses this feedback to optimize equalization parameters, resolving the circular dependency through iterative refinement.
2Measurement precision
If training bits are captured reliably, then equalization can be performed, but equalization cannot occur without reliable capture creating a circular dependency
Solution Approach 1:
The patent introduces an intermediary approach by using known test patterns and channel estimation techniques as a bridge between transmission and equalization. Instead of directly capturing training bits through the distorted channel, the system uses predetermined patterns that can be correlated with received signals to estimate channel characteristics, serving as an intermediary step that enables both reliable capture and equalization.
Solution Approach 2:
The system performs preliminary channel estimation using known test patterns before attempting to capture actual training bits. This preliminary action establishes initial equalizer settings that enable reliable capture of subsequent training data, breaking the circular dependency by preparing the equalization system in advance.
3Speed
If high data rate transmission is implemented over backplanes, then communication speed is improved, but signal distortions increase requiring complex equalization
Solution Approach 1:
The patent applies preliminary action by pre-emphasizing the transmitted signal based on predicted channel characteristics at high data rates. The system uses known test patterns to estimate channel distortion effects in advance and pre-adjusts the transmitted signal to compensate for expected distortions, reducing the burden on receiver equalization.
Solution Approach 2:
The system implements feedback-based equalization where receiver estimates of channel distortion at high data rates are fed back to adjust transmitter pre-emphasis settings. This continuous feedback loop optimizes the balance between transmission speed and signal quality by adapting equalization parameters to actual channel conditions.
Data Source
AI summary
In training a SERDES, a Common Electrical Interface (CEI) training frame, having certain bits of information embedded therein, is transmitted over a path which comprises transmitter, channel, and receiver components. The present invention analyzes the resulting received signal and determines the effective aggregate channel impulse response of these three components. The invention then determines an estimate of the inverse of this aggregate channel and uses this determination to reduce distortions that have been introduced into a signal that has been transmitted over the path.