Receiver Equalizer Training Using Channel Slope for Fast Linkup
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing PCIe link training methods are slow and inefficient, requiring 10 seconds or more for equalizer training due to strict latency limitations and frequent retransmissions at high symbol rates, which hinder fast linkup times in high-bandwidth applications.
Innovation Solution
Implement a continuous time linear equalizer with high-frequency boost, an analog-to-digital converter, and a channel estimation module to measure spectral density, using the difference in spectral densities to set operating parameters and retrieve optimized settings from a lookup table for rapid link training.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a step-by-step parameter-by-parameter optimization process is used for equalizer training, then the equalization accuracy is improved, but the training time increases to 10 seconds or more
Solution Approach 1:
The patent pre-calculates and stores optimal equalizer parameters in lookup tables before actual operation. During training, the system measures channel characteristics and directly retrieves pre-optimized parameters from the lookup table, avoiding time-consuming iterative optimization while maintaining equalization accuracy.
Solution Approach 2:
The patent creates a simplified model of the complex optimization process by pre-computing optimal parameters for various channel conditions and storing them as lookup tables. This copy of the optimization results allows rapid parameter selection without repeating the full optimization algorithm during actual training.
2Productivity
If high symbol rates are used to increase data rate, then the bandwidth is improved, but noise and signal integrity issues increase causing frequent retransmissions
Solution Approach 1:
The system performs preliminary channel characterization measurements before data transmission begins. Based on these measurements, optimal equalizer parameters are selected from lookup tables to pre-compensate for channel impairments, enabling high symbol rates with reduced retransmissions.
Solution Approach 2:
The system continuously monitors channel conditions and uses feedback to select appropriate equalizer parameters from lookup tables. This adaptive feedback mechanism maintains signal integrity at high data rates by dynamically adjusting equalization based on measured channel characteristics.
3Loss of time
If the receiver enters normal operating mode quickly after reset, then the linkup time is reduced, but the equalizer training may be incomplete
Solution Approach 1:
Optimal equalizer parameters are pre-calculated and stored in lookup tables during manufacturing or initialization. During rapid linkup after reset, the system performs minimal channel measurements and directly retrieves pre-optimized parameters, achieving both fast linkup and complete equalization.
Solution Approach 2:
The system uses pre-computed parameter sets from lookup tables that represent complete equalization solutions for various channel conditions. By copying these pre-optimized parameters rather than performing full iterative training, the system achieves rapid linkup with complete equalization quality.
Data Source
AI summary
To expedite training and achieve fast linkup times, one illustrative receiver includes: a continuous time linear equalizer configured to provide a controllable amount of high frequency boost to an analog receive signal; an analog to digital converter configured to convert the analog receive signal into a digitized receive signal; a channel estimation module configured to obtain a spectral density measurement of the digitized receive signal at each of two different frequencies, the spectral density measurements having a difference indicating a slope of a channel response; and a microcontroller configured to set at least one operating parameter of the continuous time linear equalizer based on said difference. An illustrative method includes: estimating a slope of frequency dependence of a channel response; using the slope to retrieve a set of equalization parameters from a look-up table; and employing the set of equalization parameters to exit a receiver training mode.