Optical Interconnect System Using Nonlinear Precoding and Equalization

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

The bandwidth limitation of optoelectronic devices in communication systems hinders the ability to meet increasing capacity demands, particularly in high-speed short-range fiber optic links, leading to deteriorated bit error ratio (BER) performance due to signal damage from high-order PAM signals.

Innovation Solution

An optical interconnect system incorporating a nonlinear differential precoding module, a generalized Tomlinson-Harashima precoding module, a faster than Nyquist module, a signal transmission module, a finite impulse response equalizer (FFE), and a simplified 2D constellation distortion (S2DCD) module, which perform nonlinear differential precoding, pre-equalization, high-frequency truncation filtering, strong equalization, and constellation reshaping to mitigate inter-code interference and color noise.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If high-order PAM signals (PAM-6, PAM-8) are used to enhance communication capacity, then the communication capacity increases, but the bit error ratio performance deteriorates due to system damage to high-level signals

Engineering Contradiction:
Improvecommunication capacityVSAvoidbit error ratio performance
Core Design Contradiction:
ProductivityVSReliability

Solution Approach 1:

The patent changes the signal level parameter from conventional PAM-4 to higher-order PAM-6 or PAM-8 signals, thereby increasing communication capacity. Simultaneously, it introduces nonlinear equalization algorithms to compensate for signal distortion and maintain acceptable bit error ratio performance despite the increased system damage to high-level signals.

Inventive Principle:
Principle #35Parameter changes

2Productivity

If the baud rate is increased to satisfy increasing capacity demand, then the communication capacity increases, but the system performance deteriorates due to bandwidth limitation of optoelectronic devices

Engineering Contradiction:
Improvecommunication capacityVSAvoidsystem performance
Core Design Contradiction:
ProductivityVSReliability

Solution Approach 1:

The patent applies preliminary action by implementing precoding techniques before signal transmission to pre-compensate for expected signal distortion caused by bandwidth limitations. This includes using nonlinear precoding algorithms that anticipate and counteract the frequency-dependent attenuation that will occur during transmission through optoelectronic devices with limited bandwidth.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent employs feedback mechanisms through equalization algorithms that use received signal information to adjust and compensate for distortion. The equalizer continuously adapts its parameters based on the actual signal quality observed, allowing the system to maintain performance despite bandwidth limitations by dynamically correcting for frequency-dependent losses.

Inventive Principle:
Principle #23Feedback

Data Source

PatentUS12143250B2Optical interconnect system and method for data center
Publication Date: 2024.11.12 SUZHOU UNIV
  • US12143250B2 patent drawing
  • US12143250B2 patent drawing
  • US12143250B2 patent drawing

AI summary

The invention provides an optical interconnect system and method for a data center. A nonlinear differential precoding module performs nonlinear differential precoding on an inputted original signal, to obtain a precoded signal with an increased quantity of levels. A generalized Tomlinson-Harashima precoding (GTHP) module pre-equalizes the precoded signal, to obtain a pre-equalized signal with scattered distribution. A faster than Nyquist (FTN) module performs high-frequency truncation filtering on the pre-equalized signal, to obtain a discrete signal. A signal transmission module transmits the discrete signal from a transmitting end to a receiving end. A feed-forward equalizer (FFE) performs strong equalization on the discrete signal to obtain a level slice signal, and decodes the level slice signal according to a GTHP decoding table, to obtain a decoded signal. A simplified 2D constellation distortion module processes the decoded signal to reshape a constellation, and obtains a restored signal according to a time interleaving method.