# US 2021/0152403 A1 · Adaptive equalizer, optical receiver, and adaptive equalization method

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2021/0152403 A1
- Document ID: US20210152403A1
- Status: Published application
- Kind: A1
- Application number: 17/073,716
- Filed: 2020-10-19
- Published: 2021-05-20
- Earliest claimed: 2019-11-19
- Assignees: Fujitsu Limited
- Applicants: Fujitsu Limited
- Inventors: Daisuke SASAKI; Hisao Nakashima; Nobukazu KOIZUMI
- Web page: https://patentgrep.com/patent/US20210152403A1

## CPC

- H04L 25/03178: {Arrangements involving sequence estimation techniques}
- H04L 25/14: Channel dividing arrangements {, i.e. in which a single bit stream is divided between several baseband channels and reassembled at the receiver}
- H04B 10/25073: {using spectral equalisation, e.g. spectral filtering}
- H04B 10/616: {Details of the electronic signal processing in coherent optical receivers}
- H04L 25/03057: {with a recursive structure (H04L25/03031 takes precedence)}
- H04L 25/0314: {using fractionally spaced delay lines or combinations of fractionally integrally spaced taps}
- H04L: TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION (arrangements common to telegraphic and telephonic communication H04M)
- H04B: TRANSMISSION

## Abstract

An adaptive equalizer, includes a sample buffer; and a processor coupled to the sample buffer and configured to: perform an adaptive equalization on data which has been fractionally sampled at a sampling rate higher than once a symbol rate and lower than twice the symbol rate, determine an initial value of a tap coefficient of the adaptive equalizer by using a training sequence inserted in the data, shift, by a predetermined shift amount, a sample point of one pattern from among two consecutive patterns included in the training sequence, specify a position of the training sequence in the data by replacing an original sample value with a sample value at the shifted sample point, and update the initial value of the tap coefficient based on the specified training sequence.

## Claims

### Claim 1

An adaptive equalizer, comprising:
a sample buffer; and
a processor coupled to the sample buffer and configured to:
perform an adaptive equalization on data which has been fractionally sampled at a sampling rate higher than once a symbol rate and lower than twice the symbol rate,
determine an initial value of a tap coefficient of the adaptive equalizer by using a training sequence inserted in the data,
shift, by a predetermined shift amount, a sample point of one pattern from among two consecutive patterns included in the training sequence,
specify a position of the training sequence in the data by replacing an original sample value with a sample value at the shifted sample point, and
update the initial value of the tap coefficient based on the specified training sequence.

### Claim 2 (depends on 1)

The adaptive equalizer according to claim 1, wherein
the predetermined shift amount is uniquely defined by the sampling rate and a symbol length of the one pattern.

### Claim 3 (depends on 2)

The adaptive equalizer according to claim 2, wherein
the predetermined shift amount is zero when a product of the sampling rate and the symbol length is an integer.

### Claim 4 (depends on 1)

The adaptive equalizer according to claim 1, wherein
the sample buffer that sequentially stores the data that has been fractionally sampled; and
the processor is configured to:
read sample data in a predetermined range while a selection position for the buffer is moved by one sample at a time, and
determine, as a start position of the training sequence, a buffer selection position that maximizes a correlation value between the two consecutive patterns.

### Claim 5 (depends on 1)

The adaptive equalizer according to claim 1, wherein
the sample buffer that sequentially stores the data that has been fractionally sampled; and
the processor is configured to:
when the sampling rate is f and the symbol length of the one pattern is T, calculate a correlation value between a first sample set that contains (f×T) pieces and a second sample set that contains (f×T) pieces and is subsequent to the first sample set, and
determine a buffer selection position when the correlation value is maximized, as a start position of the training sequence, based on a sample value after the phase shift.

### Claim 6 (depends on 1)

The adaptive equalizer according to claim 1, wherein
the training sequence includes m (m is a natural number) pattern pairs, each pattern pair has two same code patterns, and different pattern pairs have different code patterns.

### Claim 7 (depends on 1)

An optical receiver comprising:
an optoelectric conversion circuit that converts an optical signal received through a transmission line into an analog electrical signal;
an analog-to-digital converter that fractionally samples the analog electrical signal at the sampling rate; and
the adaptive equalizer according to claim 1, wherein the adaptive equalizer carrying out adaptive equalization on the data that has been fractionally sampled.

### Claim 8

An adaptive equalization method executed by an adaptive equalizer comprising:
performing an adaptive equalization on data which has been fractionally sampled at a sampling rate higher than once a symbol rate and lower than twice the symbol rate;
determining an initial value of a tap coefficient of the adaptive equalizer by using a training sequence inserted in the data;
shifting, by a predetermined shift amount, a sample point of one pattern from among two consecutive patterns included in the training sequence;
specifying a position of the training sequence in the data by replacing an original sample value with a sample value at the shifted sample point; and
updating the initial value of the tap coefficient based on the specified training sequence.
