KR20030029199A - Apparatus and method for time domain echo canceller of DMT - Google Patents

Abstract

The present invention is to provide a time domain echo canceller device of the DMT and a control method thereof, comprising: an DMT echo having an encoder, a first IFFT, a P / S converter, a transmit / receive hybrid filter, a first delay unit, and a third adder A canceller, comprising: a second delay unit for delaying a signal of the transmission / reception hybrid filter; A first adder which adds an output of the second delay unit and an output of a fourth adder; A third delay unit for delaying the output of the first adder; An S / P converter converting serial data of the third delay unit into parallel data; An FFT for FFTing the output of the S / P converter; A second adder which subtracts the output of the echo canceller from the output of the FFT; An echo canceller configured to receive an output of the second adder and to perform an echo canceling operation by receiving an output of the encoder; Second and third IFFTs respectively performing an IFFT on an output of the echo canceller; A signal compensator configured to receive the outputs of the third adder and the second IFFT and perform signal compensation; By including the signal compensator and the fourth adder for adding the output of the third IFFT and output to the first adder, it is possible to improve the performance of the echo canceller by obtaining the echo impulse response in the time domain on the CO side Will be.

Description

DMT's time domain echo canceller device and its control method {Apparatus and method for time domain echo canceller of DMT}

The present invention relates to a time domain echo canceller of Discrete Multitone Modulation (DMT). In particular, an echo impulse response of a time domain is obtained from a central office (CO) to obtain echo cancellation. The present invention relates to a time-domain echo canceller device of a DMT and a control method thereof, which are suitable for improving the performance of a circulator.

In general, DMT is an American National Standard Institute (ANSI) standard modulation technology used for Asynchronous Digital Subscriber Line (ADSL).

1 is a block diagram of a time domain echo canceller of a conventional DMT.

Here, reference numeral 11 is an encoder for encoding input parallel data, and 12 is a first IFFT for performing an inverse fast fourier transform (IFFT) on a signal encoded by the encoder 11 to send a reverb signal. 13 is a P / S (Parallel to Serial) converter for converting parallel data output from the first IFFT 12 into serial data, and 14 is an input of the P / S converter 13. It is a transmission / reception hybrid filter that transmits and receives data and receives downsamples a remote signal through a line.

Reference numeral 15 denotes a first adder which adds an output of the transmission / reception hybrid filter 14 and an output of the signal compensator 24, and 16 denotes parallel data of serial data output from the first adder 15. S / P (Serial to Parallel) transform unit to convert, 17 is an FFT performing a Fast Fourier Transform (FFT) to the output of the S / P conversion unit 16, 18 is an output of the FFT (17) Is a second adder subtracting the output of the echo canceller 20, and 19 is a receiver that receives the output of the second adder 18.

In addition, reference numeral 20 denotes an echo canceler that receives an output of the second adder 18 and an output of the encoder 11 to perform echo canceling, and 21 denotes an echo canceller ( A second IFFT performing an IFFT on the output of 20), 22 is a delay unit for delaying the output of the first IFFT 12, and 23 is a delay unit 22 at the output of the first IFFT 12 And a third adder subtracting the output of the reference signal, and 24 denotes an output of the third adder 23 and the second IFFT 21 by performing signal compensation to the second adder 15. It is a signal cancellation unit (Tail Cancellation and Cyclic Reconstruction).

Therefore, the echo canceller of the DMT includes the first to third adders 15, 18, 23, the S / P converter 16, the FFT 17, the echo canceller 20, and the second IFFT 21. ), A delay unit 22 and a signal compensator 24.

In the DMT, the signal sent by the transmitting end is incomplete by the transmitting / receiving hybrid filter 14 so that an unwanted signal enters the receiving end.

You want to receive only remote signals at the receiving end, but you will receive unwanted signals, ie echoes sent from the transmission.

In general, the echo path loss (Kep) is 17dB, the channel loss (Channel Loss, Kch) is 23dB, and the signal-to-noise ratio (SNR) margin is 44dB.

Therefore, ECHO (Kec) to be removed at the receiving end is as follows.

Kec = SNR + Kch-Kep = 50 dB

That is, it should be eliminated more than 50dB.

However, since the DMT sends symbols in units of frames, there are discontinuous parts in time domain echoes.

In this part, there is a tail by the previous symbol and a transition by the current symbol.

In addition, in DMT, the received frame is matched to the remote signal, so there is a frame mismatch between the echo and the remote signal.

In the prior art, the echo impulse response obtained in the time domain is divided into three cases to compensate for such discontinuities, and even frame mismatch with the remote signal, and then FFT to perform echo cancellation in the frequency domain.

2 is a flowchart illustrating a time domain echo canceller control method of a conventional DMT.

As shown therein, a step of modeling an echo path through a reverb signal is performed and IFFT to obtain an echo impulse response (ST11); Determining a valid portion of the obtained echo impulse response, determining where the discontinuous portion of the echo signal is based on its position, and determining the degree of inconsistency of the remote signal of the echo (ST12) (ST13); Performing a convolution in the time domain after the determination to compensate for an inconsistency (ST14); After the compensation, an FFT is performed and an echo canceller is performed in the frequency domain (ST15).

Therefore, in the prior art, the echo path is modeled in the frequency domain through a reverb signal (PATTERN DATA) and then IFFT is obtained to obtain an echo impulse response.

From the obtained echo impulse response, a valid section is determined, where the discontinuous portion of the echo signal is determined according to its position, and the mismatch between the echo and the remote signal is determined.

Once the discrete portion and the degree of mismatch are determined, convolution is performed in the time domain with the selected portion of the time domain signal at the transmitting side to compensate for this portion.

This convolution compensates for (periodically) discontinuous parts (ie, TAIL of the previous symbol + TRANSITION of the current symbol), and also compensates for mismatched parts of the remote signal and echo.

If fully compensated, the echo has no transitions and tails, and the remote signal and frame sync are correctly matched.

FFT is performed to perform echo cancellation in the frequency domain.

Referring to the operation of the prior art in more detail as follows.

The sender first sends a signal (periodic signal), which enters the receiver via an echo path.

This signal is called an echo, and the echo canceller 20 models a complex one tap for each bin to model the transmission / reception hybrid filter 14 that is an echo path. At this time, the update method uses LMS (Least Mean Squared).

After the echo path is modeled, a time domain echo impulse response is obtained after IFFT through the second IFFT 21.

Since the DMT sends data in units of frames, the transition of the tail of the previous symbol and the current symbol occurs. Also, because the frame is tuned to the remote signal, the echo signal is out of frame.

Thus, the signal compensator 24 and the first adder 15 are blocks that compensate for tails, transitions, and frame mismatches.

According to the shape of the time domain echo impulse response obtained as described above, the tail of the previous symbol, the transition of the current symbol, and the frame mismatch of the first IFFT 12 of the transmitting side in the signal compensation unit 24. The output is compensated with the value obtained from the convolution with the selected sample during the time-domain echo impulse response.

This is FFTed by the FFT 17 and finally eliminated by the second adder 18. At this time, of course, the section in which the most energy is collected among the 512 samples is taken.

3 is a waveform diagram showing a form of an echo impulse response when performing a time domain echo canceller of a conventional DMT.

Therefore, there are three cases as shown in (a)), (b) and (c) of FIG. 3 according to the type of echo impulse response.

However, this prior art has a problem that the configuration is very complicated because it has to be configured to satisfy the operation for the three cases as shown in FIG. That is, the values that need to be convolved in the signal compensator 24 are different according to the frame mismatch, thereby increasing the complexity of the signal compensator 24.

In addition, there is a problem in that the performance of echo canceling occurs in the process of compensating the transition and the non-transition part to match the frame synchronization of the echo and the remote signal received at the receiver. That is, the number of samples of the echo impulse response is actually 512 samples, of which about 256 samples are written, and there is a problem in that accuracy is reduced to compensate for frame mismatch with 256 samples.

In addition, since the FFT eliminates the compensated signal in the frequency domain, a truncation error that may occur in the FFT may be included.

Accordingly, the present invention has been proposed to solve the conventional problems as described above, and an object of the present invention is to obtain a time-domain echo impulse response on the CO side, and to improve the performance of the echo canceller. It is to provide a canceller device and a control method thereof.

In order to achieve the above object, the DMT time domain echo canceller device according to the embodiment of the present invention,

A DMT echo canceller comprising an encoder, a first IFFT, a P / S converter, a transmit / receive hybrid filter, a first delay unit, and a third adder, comprising: a second delay unit for delaying a signal of the transmit / receive hybrid filter; A first adder which adds an output of the second delay unit and an output of a fourth adder; A third delay unit for delaying the output of the first adder; An S / P converter converting serial data of the third delay unit into parallel data; An FFT for FFTing the output of the S / P converter; A second adder which subtracts the output of the echo canceller from the output of the FFT; An echo canceller configured to receive an output of the second adder and to perform an echo canceling operation by receiving an output of the encoder; Second and third IFFTs respectively performing an IFFT on an output of the echo canceller; A signal compensator configured to receive the outputs of the third adder and the second IFFT and perform signal compensation; And a fourth adder which adds an output of the signal compensator and the third IFFT to output the first adder.

In order to achieve the above object, a time domain echo canceller control method of a DMT according to an embodiment of the present invention,

Sending a reverb signal to perform an FFT on the echo and modeling the frequency domain echo path as a complex one tap in the frequency domain; A second step of obtaining a time domain echo impulse response by IFFTing the frequency domain echo path after the first step, and determining a valid section among the time domain echo impulse responses; A third step of delaying the receive path after the second step and then compensating with the back of the previous symbol and the back of the current symbol to form an echo signal; A fourth step of canceling echo in the time domain after the third step; And a fifth step of delaying again so that the delay of the reception path is a symbol delay after the fourth step.

1 is a block diagram of a time domain echo canceller of a conventional DMT.

2 is a flowchart illustrating a time domain echo canceller control method of a conventional DMT.

3 is a waveform diagram showing a form of an echo impulse response when performing a time domain echo canceller of a conventional DMT.

4 is a block diagram of a DMT time domain echo canceller device according to the present invention.

5 is a flowchart illustrating a method of controlling a time domain echo canceller of a DMT according to the present invention.

FIG. 6 is a waveform diagram illustrating a shape of a time domain echo impulse response when a DMT performs time domain echo canceller according to the present invention.

FIG. 7 is a waveform diagram after giving a delay of a predetermined sample with respect to FIG. 6 when performing time-domain echo canceller of DMT according to the present invention.

Explanation of symbols on the main parts of the drawings

11: encoder 12, 21, 33, 34: IFFT

13: P / S converter 14: transmission and reception hybrid filter

15, 18, 23, 36: Adder 16: S / P Converter

17: FFT 19: Receiver

20: echo canceller 22, 31, 32: delay unit

35: signal compensation unit

Hereinafter, an embodiment according to the present invention configured as described above, a time domain echo canceller device of the DMT and a control method thereof will be described in detail with reference to the accompanying drawings.

4 is a block diagram of a DMT time domain echo canceller device according to the present invention.

As shown therein, the encoder 11, the first IFFT 12, the P / S converter 13, the transmit / receive hybrid filter 14, the first delay unit 22, and the third adder 23 A DMT echo canceller comprising: a second delay unit (31) for delaying a signal of the transmission / reception hybrid filter (14); A first adder (15) for adding the output of the second delay unit (31) and the output of the fourth adder (36); A third delay unit (32) for delaying the output of the first adder (15); An S / P converter (16) for converting serial data of the third delay unit (32) into parallel data; An FFT (17) for FFTing the output of the S / P converter (16); A second adder (18) which subtracts the output of the echo canceller (20) from the output of the FFT (17); An echo canceler (20) for receiving an output of the second adder (18) and receiving an output of the encoder (11) to perform echo cancellation; Second and third IFFTs (33) (34) which perform IFFT on the output of the echo canceller (20), respectively; A signal cancellation unit (35) for performing signal compensation upon receiving the outputs of the third adder (23) and the second IFFT (33); And a fourth adder 36 that adds the outputs of the signal compensator 35 and the third IFFT 34 to the first adder 15.

5 is a flowchart illustrating a method of controlling a time domain echo canceller of a DMT according to the present invention.

As shown therein, a first step (ST21 to ST23) of performing a FFT on an echo by sending a reverb signal and modeling a frequency domain echo path as a complex one tap in the frequency domain; A second step (ST24) (ST25) of obtaining a time domain echo impulse response by IFFTing the frequency domain echo path after the first step, and determining a valid section among the time domain echo impulse responses; A third step (ST26) (ST27) of delaying the reception path after the second step and compensating with the latter part of the previous symbol and the latter part of the current symbol and forming an echo signal; A fourth step ST28 of eliminating echo in the time domain after the third step; And a fifth step (ST29) for delaying again so that the delay of the reception path is a symbol delay after the fourth step.

In the fourth step, the echo signal is multiplied by the frequency domain echo path and the frequency symbol sent from the transmitter, and then subtracted by 512 point IFFT to remove the echo in the time domain.

The operation of the DMT time domain echo canceller device and its control method according to the present invention configured as described above will be described in detail with reference to the accompanying drawings.

In the training process, the transmitting side sends a reverb signal, and the echo enters the receiving path through the transmit / receive hybrid filter 14.

Of course, it sends a reverb signal, so there is no reason to worry about echo frames in the receive path.

The incoming echoes are then FFTed at the FFT 17 to model the frequency domain echo paths as complex one taps in the frequency domain.

And the Complex One tab obtained during the training is fixed. This is because it does not update in data mode.

IFFT the well-modeled frequency-domain echo path to obtain the time-domain echo impulse response.

A valid section is determined during the time-domain echo impulse response. The standard used to determine the interval M where the power is most concentrated is generally 200 samples when 512 points IFFT.

This delays the receive path so that the first start of M is at the very beginning of the time-domain impulse response.

FIG. 6 is a waveform diagram illustrating a time domain echo impulse response when a DMT performs a time domain echo canceller according to the present invention, and FIG. 7 is a constant diagram of FIG. 6 when performing a time domain echo canceller according to the present invention. This is a waveform diagram after giving a delay as much as a sample.

So, for example, if the shape of the time-domain echo impulse response is as shown in Fig. 6, it is made as shown in Fig. 7 after giving a delay of x / 8 samples to the receiving path.

The portion to be compensated is the transition of the tail of the previous symbol and the current symbol, which is compensated with 199 samples after the previous symbol (but M = 200) and 199 samples after the current symbol.

This results in an echo signal without tails and transitions.

If the tail and transition-free echo signals are multiplied by the frequency domain echo path and the frequency symbol sent by the transmitter, and subtracted with 512 POINTS IFFT, all echoes are lost in the time domain.

What is important here is that the IFFT is multiplied in the frequency domain, but the values convolved in the time domain are the same.

After eliminating all the echoes, if the delay of the reception path is delayed again so that the symbol delay is completed, the time domain echo canceller is completed.

Thus, the received frame synchronization that is set to the remote signal does not deviate at all and becomes an echo canceller in the time domain.

Referring to the operation of the present invention in more detail as follows.

First, the echo canceller of the DMT of the present invention includes the first to fourth adders 15, 18, 23, 36, the S / P converter 16, the FFT 17, and the echo canceller 20. And second and third IFFTs 33 and 34, first to third delay units 22 and 31 and 32, and a signal compensator 24.

So the first IFFT 12 on the sending side sends a signal (periodic signal) and it enters the receiving end via the echo path.

The incoming signal is called an echo, and the echo canceler 20 models a complex one tap for each bin to model the transmission / reception hybrid filter 14 which is an echo path. At this time, the update method uses LMS.

If the echo path is modeled, a time domain echo impulse response is obtained after IFFT through the second IFFT 33.

After observing the time domain echo impulse response, the number of delays of the second delay unit 31 is determined. This is to virtually match the frame of the echo and remote signal.

Since the delay is delayed by the second delay unit 31, only a predetermined portion of the IFFT of the transmitting side is taken to model the tail of the previous symbol and the transition of the current symbol, and then the time compensator 35 performs a time domain echo impulse response and the like. The convolution is performed and brought to the fourth adder 36.

In the third IFFT 34, the frequency domain symbol (complex value) and the frequency domain echo response (complex value) of the transmitter are multiplied to compensate for a steady state portion excluding the tail of the previous symbol and the transition of the current symbol. Then IFFT.

This is because the two values can be multiplied in the time domain and multiplied in the frequency domain, then IFFTed to obtain the same value that is sent to the time domain.

By doing this, you get much less complexity than getting the value through convolution.

In addition, unlike the prior art, the echo is compensated in the time domain and not in the frequency domain, but the echo is eliminated in the time domain.

The delay delayed by the second delay unit 31 to the reception path is arbitrarily compensated by the third delay unit 32, and thus does not affect the reception path.

Because the delay of the second delay unit 31 + the delay of the third delay unit = symbol delay does not affect the reception path.

Unlike the prior art, since the second delay unit 31 gives a delay, the IFFT output of the transmitting side is brought to convolution for compensating the tail and the transition, which always brings a value of a constant portion. Accordingly, the complexity of the signal compensator 35 that needs to perform convolution is significantly reduced.

In addition, instead of using convolution to compensate for the steady state, multiply in the frequency domain and then IFFT the third IFFT 34, thereby improving performance and reducing complexity in convolution.

In addition, unlike the related art, since the truncation error is eliminated in the time domain without the FFT to eliminate the echo in the frequency domain, truncation error can be prevented.

As such, the present invention improves the performance of the echo canceller by obtaining the echo impulse response in the time domain on the CO side.

Although the preferred embodiment of the present invention has been described above, the present invention may use various changes, modifications, and equivalents. It is clear that the present invention can be applied in the same manner by appropriately modifying the above embodiments. Accordingly, the above description does not limit the scope of the invention as defined by the limitations of the following claims.

As described above, the DMT time domain echo canceller apparatus and control method thereof have an effect of obtaining an echo impulse response in the time domain on the CO side to improve the performance of the echo canceller.

In addition, while the prior art has a very complicated complexity of convolving the previous symbol and the current symbol according to the shape of the time-domain echo impulse response as shown in FIG. 3, the present invention delays the reception path. By eliminating the tail, the latter part of the previous symbol and the current symbol's definite part of the current symbol are removed, which significantly reduces the complexity in convolution, thus simplifying the algorithm of the entire system. There is an effect that can be implemented.

In addition, in contrast to the conventional use of convolution in the time domain to compensate for the steady state, the present invention also has the effect of using a 512 point IFFT to reduce the complexity and improve the performance of the echo canceller.

Furthermore, the present invention has an effect of preventing truncation errors since all of them are removed in the time domain without FFT to eliminate echoes in the frequency domain.

Claims (3)

In the DMT echo canceller having an encoder, a first IFFT, a P / S converter, a transmit / receive hybrid filter, a first delay unit, and a third adder, A second delay unit delaying a signal of the transmission / reception hybrid filter; A first adder which adds an output of the second delay unit and an output of a fourth adder; A third delay unit for delaying the output of the first adder; An S / P converter converting serial data of the third delay unit into parallel data; An FFT for FFTing the output of the S / P converter; A second adder which subtracts the output of the echo canceller from the output of the FFT; An echo canceller configured to receive an output of the second adder and to perform an echo canceling operation by receiving an output of the encoder; Second and third IFFTs respectively performing an IFFT on an output of the echo canceller; A signal compensator configured to receive the outputs of the third adder and the second IFFT and perform signal compensation; And a fourth adder configured to add the signal compensator and the outputs of the third IFFT to the first adder to output the signal to the first adder. Sending a reverb signal to perform an FFT on the echo and modeling the frequency domain echo path as a complex one tap in the frequency domain; A second step of obtaining a time domain echo impulse response by IFFTing the frequency domain echo path after the first step, and determining a valid section among the time domain echo impulse responses; A third step of delaying the receive path after the second step and then compensating with the back of the previous symbol and the back of the current symbol to form an echo signal; A fourth step of canceling echo in the time domain after the third step; And a fifth step of delaying again so that the delay of the reception path becomes a symbol delay after the fourth step. The method of claim 2, wherein the fourth step, A method of controlling the time domain echo canceller of DMT, characterized in that the echo signal is removed from the time domain by multiplying the frequency domain echo path from the echo signal by the frequency symbol sent from the transmitter and subtracting it by a 512 point IFFT.