WO2015123989A1

WO2015123989A1 - Channel equalization method and system based on time-domain channel estimation

Info

Publication number: WO2015123989A1
Application number: PCT/CN2014/086212
Authority: WO
Inventors: 张帆; 王丹; 丁瑞
Original assignee: 北京大学
Priority date: 2014-02-24
Filing date: 2014-09-10
Publication date: 2015-08-27
Also published as: CN103812806B; CN103812806A

Abstract

Disclosed are a channel equalization method and system based on time-domain channel estimation. The method comprises: firstly, inserting a training sequence into a transmission end signal; then, performing channel equalization of the received training sequence part on a reception end by using a filter, and updating a filter coefficient; and filtering a data signal in a time domain by using the filter updated in the previous step, so as to implement channel equalization. In the present invention, channel equalization is performed by using a training sequence and a time-domain filter. A frame structure of a training sequence part is formed by M sequences and 0 sequences in an interleave manner, which helps the time-domain filter equalize polarization crosstalk a signal meets, so as to unlock polarization multiplexing; in addition, a coefficient of the time-domain filter is updated by using an RLS algorithm, the algorithm complexity is low, computation time can be saved, the sensitivity to a modulation format is absent, and the applicability is wide.

Description

Channel equalization method and system based on time domain channel estimation

Technical field

The present invention relates to the field of coherent optical communication transmission, and relates to a channel equalization method based on time domain channel estimation, and a system for implementing the method.

Background technique

Coherent communication based on digital signal processing technology is an important solution for long-distance optical fiber communication transmission systems. How to overcome the communication channel damage is an important issue to be solved. Communication channel impairments reduce the signal to noise ratio of the signal and introduce intersymbol interference, resulting in the generation of errors. In fiber-optic communication systems, an equalization algorithm estimates the channel and compensates for linear distortion of the channel to attenuate or eliminate inter-symbol interference.

At present, there are two main types of equalization algorithms, one is time domain equalization (TDE) and the other is frequency domain equalization (FDE). Both types of algorithms have the same equalization effect.

1) Time Domain Equalization (TDE) algorithm. The algorithm performs channel impulse response estimation based on the training sequence or signal-based constellation characteristics, and compensates the signal in the time domain. The channel estimation algorithm based on the training sequence is more complicated. The channel estimation algorithm based on the characteristics of the signal constellation depends on the modulation format. When the modulation format is changed, the algorithm also needs to be changed, and the algorithm effect is different. Due to the dispersion in the fiber, when the dispersion is large, the equalization algorithm requires more equalizer taps, resulting in higher algorithm complexity.

2) Frequency Domain Equalization (FDE) algorithm. The algorithm performs estimation of the channel transfer function based on the training sequence and compensates the signal in the frequency domain. The channel estimation algorithm is simple, and when the dispersion is large, the algorithm complexity is still low. But the algorithm needs to insert a cyclic prefix/suffix to the signal, which causes a certain signal rate overhead.

Summary of the invention

The present invention provides an effective channel equalization method based on time domain channel estimation, and a system for implementing the method.

To achieve the above object, the present invention adopts the following technical solutions:

A channel equalization method based on time domain channel estimation includes the following steps:

The first step: inserting a training sequence with flat spectral characteristics into the signal at the transmitting end;

The second step: at the receiving end, after the front-end signal processing, the received training sequence is filtered by the FIR filter to achieve channel equalization. The FIR filter coefficients are updated using the RLS algorithm criteria.

The third step: using the time domain FIR filter obtained in the second step to filter the data signal in the time domain to achieve channel equalization.

Further, the receiving end first performs front end data processing on the received signal, including dispersion coarse compensation and carrier frequency. Rate recovery, receive matched filtering, and digital synchronization; and after the third step, back-end data processing, including carrier phase recovery.

Further, the signal at the transmitting end is frame-transmitted. As shown in FIG. 2, the frame structure of each frame includes two polarization directions, each of which includes a training sequence and a data signal. The training sequence in the X polarization direction is the sequence t ₁ arranged in time in the period, and the training sequence in the Y polarization direction is the sequence t ₂ which is periodically arranged in time, and:

t ₁ =(t _x ,0),t ₂ =(0,t _y ),t _x =t _y ,

Wherein t _x and t _y are M sequences, and "0" represents a sequence having a length equal to 0 in length and M sequence.

Further, the selection of the M sequence length is determined by the channel carrier frequency offset and the degree of phase drift; the selection of the number of M sequences is determined by the intensity of the spontaneous radiated noise in the channel; the length of the data signal (data after each frame training sequence) The choice is determined by the degree of drift of the channel transfer function.

An equalization system based on time domain channel estimation for implementing the above method, comprising a transmitting end and a receiving end,

The transmitting end includes:

a training sequence insertion module for inserting a training sequence having flat spectral characteristics into a signal at a transmitting end;

The receiving end includes:

The FIR filter uses the time domain filter tap coefficients to filter the data signal in the time domain to achieve channel equalization.

The training sequence comparison module is connected to the FIR filter module for performing a difference between the training sequence and the data signal to obtain an error value e.

The filter coefficient updating module is connected to the training sequence comparison module to update the FIR filter coefficients by using the RLS criterion according to the error value e.

Further, it also includes:

The front end data processing module is configured to preprocess the signal received from the transmitting end and then send the signal to the FIR filter module.

A back-end data processing module is coupled to the FIR filter for carrier phase recovery.

Further, the preprocessing performed by the front end data processing module includes: coarse dispersion compensation, carrier frequency recovery, receive matching filtering, and digital synchronization.

Further, a decision module is further included, and the back end data processing module is connected to restore the received signal information into binary data.

Compared with the prior art, the positive effects of the present invention are:

The present invention utilizes a training sequence and a time domain filter for channel equalization. Compared with the traditional time domain blind equalization (CMA) algorithm, the method has different frame structures and different time domain filter updating algorithms. First, the training sequence is partially between the M sequence and the 0 sequence. The interpolation structure facilitates the equalization of the polarization crosstalk received by the signal by the time domain filter, thereby unlocking the polarization multiplexing. Secondly, the RLS algorithm has better convergence than the traditional time domain blind equalization (CMA). Therefore, the present invention only uses the training sequence part to update the time domain filter coefficients, and the algorithm complexity is small, which can save computation time. The modulation format is not sensitive and has wide applicability.

DRAWINGS

1 is a flow chart of a time domain channel estimation based equalization method according to an embodiment of the present invention.

2 is a schematic diagram showing the structure of a signal frame according to an embodiment of the present invention.

FIG. 3 is a schematic diagram of filter coefficient update according to an embodiment of the present invention.

4 is a schematic diagram showing experimental results of a 400 km transmission of a 1.24 Tb/s 64QAM signal according to an embodiment of the present invention.

detailed description

The present invention will be further described in detail below through specific embodiments and the accompanying drawings; the flow of the method of the present invention is as shown in FIG.

The first step is to insert a training sequence at the front end of the signal at the transmitting end, and design according to the frame structure. Several M sequences with flat spectral characteristics are selected.

The second step: at the receiving end, after the front end signal processing, the received training sequence is filtered by the FIR filter. The filter coefficients are updated according to well-known RLS algorithm criteria.

The third step: using the time domain filter obtained in the second step to filter the data signal in the time domain to achieve channel equalization.

The fourth step: the back-end signal processing, carrier phase recovery, and decision.

RLS principle

Receiver signal processing is based on the well-known RLS (Recursive Least Squares) algorithm. The specific principles are as follows:

The FIR filter has a vector coefficient w(n) at time n and a cost function value J(n) at time n.

Let the training sequence in the data of the transmitting end be d _TS (n), the training sequence in the data of the receiving end be u _TS (n) (the angular standard TS represents the training sequence), the signal part in the data is u(n), and the filtering output training sequence is y _TS (n), the filtered output signal part is y(n)

y _TS (n)=w(n) ^T u _TS (n)

e(n)=d _TS (n)-y _TS (n)

Where λ is the forgetting factor. The effect of introducing the forgetting factor is that the error near the n-time is assigned a larger weight, and the error far from the n-time is given a smaller weight, ensuring that the observation data in a certain period of time in the past is "forgotten".

In order to get the minimum value of the cost function, you can derive the cost function.

Solutions have to:

R(n)w(n)=r(n)

among them:

According to the definition of R(n) and r(n), and after mathematical derivation, the recursive formula of w(n-1) and w(n) can be obtained. Where R(n) is the correlation matrix of the data u _TS . R(0)=σI defines the initial time correlation matrix value, where I is the identity matrix. It is generally expected that the initial value of the correlation matrix will occupy a small proportion in R(n), where σ is 0.1.

Implementation steps:

The implementation of the technical solution will be specifically described below in conjunction with the algorithm flow chart 3 of the embodiment.

The steps of updating the filter coefficients of the present time domain channel equalization scheme are as follows:

First, initialize w(0)=0; R(0)=σI;

For n = 1, 2, 3...N, where N is the number of filter updates, determined by the difference between the length of the training sequence and the length of the filter.

Calculation: y _TS (n)=w(n-1) ^T u _TS (n)

Estimation error e(n)=d _TS (n)-y _TS (n)

Define P(n)=R(n) ^-1

definition

Update

Update filter coefficients

w(n)=w(n-1)+k(n)e ^* (n)

The resulting filter coefficient after the training sequence is W=w(N).

The receiver signal u(n) is convolved with W to obtain a channel-equalized signal y(n).

As shown in FIG. 1, after the channel equalized signal, the back-end signal processing, that is, the carrier phase recovery, is performed and judged.

Figure 4 is a schematic diagram showing the experimental results of a 400km transmission of a 1.24Tb/s 64QAM signal. The horizontal axis is the fiber input power of the signal. Comparing the results of the equalization method of the present invention with a common algorithm for time domain equalization (the norm algorithm (CMA) plus cascading multimode algorithm (CMMA) plus direct decision minimum mean square error (DD_LMS)), the results show that the present invention The equalization algorithm is superior to the commonly used time domain equalization algorithm because the noise of the channel has a large influence on the filter tap estimation of the time domain equalization algorithm.

The above embodiments are only used to illustrate the technical solutions of the present invention, and the present invention is not limited thereto, and those skilled in the art can modify or replace the technical solutions of the present invention without departing from the spirit and scope of the present invention. The scope of protection shall be as stated in the claims.

Claims

A channel equalization method based on time domain channel estimation, the steps of which are:

1) inserting a training sequence into the transmitter signal;

2) performing channel equalization on the received training sequence portion by using a filter at the receiving end, and updating the filter coefficients;

3) Filter the data signal in the time domain by using the filter updated in the previous step to achieve channel equalization.
The method of claim 1 wherein said transmitting end framing the signal, the frame structure of each frame comprising two polarization directions, each polarization direction comprising a training sequence and a data signal.
The method according to claim 2, wherein said polarization directions are respectively an X polarization direction and a Y polarization direction; wherein the training sequence of the X polarization direction is a training of the sequence t 1 , Y polarization direction periodically arranged in time The sequence is a sequence t 2 arranged in time period; t 1 = (t x , 0), t 2 = (0, t y ), t x = t y , and "0" indicates that the length is equal in length to the sequence t x A sequence of values 0.
The method according to claim 3, wherein said sequence t x and sequence t y are M sequences; the length of said M sequence is determined by a channel carrier frequency offset and a degree of phase drift; The number of sequences is determined by the intensity of the spontaneous radiated noise in the channel; the length of the data signal is determined by the degree of drift of the channel transfer function.
The method of claim 1 wherein said step 2) said receiving end first performs front end data processing on the received signal, including dispersion coarse compensation, carrier frequency recovery, receive matched filtering, and digital synchronization; The received training sequence portion performs channel equalization; and after step 3) implements channel equalization, performs back-end data processing, including carrier phase recovery.
A method according to any one of claims 1 to 5, wherein said filter is an FIR filter; said receiving end updates the FIR filter coefficients using the RLS algorithm criteria.
A channel equalization system based on time domain channel estimation, comprising: a transmitting end and a receiving end, wherein the transmitting end comprises: a training sequence insertion module, configured to insert a training sequence into a signal of the transmitting end; The method includes a filter for performing channel equalization on the received training sequence portion, updating the filter coefficients, and filtering the data signal in the time domain by using filter tap coefficients to achieve channel equalization.
The system of claim 7 including a training sequence comparison module for performing a difference between the training sequence and the data signal to obtain an error value e; and a filter coefficient updating module for utilizing the RLS based on the error value e Guidelines to update the filter coefficients.
The system of claim 7 wherein said transmitting end framing the signal, the frame structure of each frame comprising two polarization directions, each polarization direction comprising a training sequence and a data signal; said filter The FIR filter; the receiving end updates the FIR filter coefficients using the RLS algorithm criteria.
The system according to claim 9, wherein said polarization directions are an X polarization direction and a Y polarization direction, respectively; wherein the training sequence of the X polarization direction is a training of a sequence t 1 , Y polarization direction periodically arranged in time. The sequence is a sequence t 2 arranged in time period; t 1 = (t x , 0), t 2 = (0, t y ), t x = t y , and "0" indicates that the length is equal in length to the sequence t x A sequence of values 0.