JPS6167330A - Echo eliminating method - Google Patents

Abstract

PURPOSE:To shorten a focussing time, by adaptively changing the correcting quantity of the tap factor of an adaptive filter in accordance with the correlative value of the code of an echo replica with the code of the difference signal between a received signal and the echo replica. CONSTITUTION:A binary data sequence from a terminal 1 is supplied to a transmitting section 3 and adaptive digital filter (ADF)6. At the transmitting section 3 the data sequence is converted into transmission line codes and sent to a two-wire transmission lien 17 through a hybrid (HYB)4. The output of a multiplier 13 is also inputted in the ADF6 and an echo replica corresponding to an echo produced at the HYB4 is generated and inputted in a subtractor 8 after D/A conversion 7. An error signal produced by subtracting the echo replica from the mixed signal of a received signal inputted through an LPF5 and the echo component of a transmitting signal is produced at the subtractor 8 and inputted in a receiving section 14 and code detector 9. Then the correlation 11 between the code detected value 10 of the echo replica and code detected value 9 of the error signal is taken and the correcting value of the tap factor of the ADF6 is updated in accordance with the correlated value during the period when transmitting signals exist only.

Description

Detailed Description of the Invention (Industrial Application Field) The present invention relates to an echo cancellation method for realizing two-wire bidirectional digital transmission. Echo cancellation is known as a common technology for realizing two-way digital transmission.
COU8TIC8,8PBECH,ANDSIGHAL
PROCESSING) Volume 27, No. 6, 1979.

(pages 768-781). If a pseudo echo (echo replica) corresponding to the transmitted data sequence is generated using the echo canceller 511 and an adaptive filter having coefficient taps equal to the length of the echo impulse response, then from K, 2 i
In the M/4-wire conversion circuit, when trying to suppress the echo leaking from the transmitting circuit to the receiving circuit,
Each tap coefficient of the adaptive filter is successively corrected by correlating the error signal obtained by subtracting the echo replica from the mixed signal of the echo and the received signal with the transmitted data.

The coefficient correction of such an adaptive filter, that is, the convergence algorithm of the echo canceller, is described in the above-mentioned references, and a representative example is the Stochastic Take Iteration algorithm (5 to chast
IC 1teration algorithm)
and the sign algorithm
m) is known. Here, echo canceller 1
When attempting to realize chipping, it is desirable to use a method that can apply digital device technology, which has recently made remarkable progress. At this time, when trying to configure the above-mentioned adaptive filter using a digital filter, an analog/digital (A/D) converter and a digital/analog (1)
/A) Requires a converter The number of pits required for the D/A converter is determined from system requirements.1 For example, in application to a subscriber line, approximately 12 bits are required. On the other hand, the required number of bits of the A/D converter depends not only on the system conditions but also on the convergence algorithm of the echo canceller described above. For example, in application to a subscriber line, if the striker-take-iteration algorithm is used, 8
The sine algorithm has the advantage of requiring only one bit, whereas the signal requires only one bit.

From the point of view of reducing the hardware scale, it is clear that adopting the latter algorithm is advantageous.

However, the sine algorithm has the disadvantage of a long convergence time.Also, when a binary code such as a bi7-factor code is used as a transmission path code, residual echo (the difference between an echo and an echo replica) ) To lower the level below the received signal level, 1) Add random noise to the input signal of the A/D converter (code detector), and adjust the amplitude of this random noise according to the received signal level. need to be controlled. Therefore, there is a drawback that the hardware scale becomes large. (Objective of the Invention) Therefore, an object of the present invention is to provide an echo removal method that requires a short convergence time and a small hardware scale.

(Structure of the Invention) According to the present invention, there is provided an echo removal method for removing an echo leaking from the transmitting side to the receiving side on the 4@ side of a 2-wire/4fil conversion circuit, and based on the transmitted signal and the received signal. In an echo removal method in which an echo replica is generated by adaptive filtering processing and the echo is removed by subtracting the echo replica from a signal in which a received signal and the echo are mixed (mixed signal), the transmitted signal is not a non-signal. During the period, the sign of the difference signal between the mixed signal and the echo replica signal is correlated with the sign of the echo replica, and the tap coefficient correction digits in the adaptive filtering process are successively updated according to the correlation value; There is obtained an echo cancellation method characterized in that the modification of the tap coefficient is stopped during a period in which there is no transmitted signal.

(Principle of the Invention) The present invention provides a correlation value between the code of an echo replica and the error signal (=echo echo replica + received signal).
Focusing on the fact that the residual echo (=echo-echo replica) changes depending on the level, convergence is achieved by adaptively changing the correction amount of the tap coefficient of the adaptive filter according to the magnitude of the correlation value. Enables time savings. At this time, in the no-signal period of the transmitted signal, which is generated in a burst pattern (10-hour periods with signal and 10-hour periods without signal), the error signal includes both echoes and echo replicas. Therefore, since the sign of the error signal, which is one input of the correlator, matches the sign of the received signal, a problem arises in that the desired correlation value cannot be obtained. By using the timing signal of the no-signal interval generated by the transmitting side control unit which has the function of generating various timing signals to convert the transmitted signal into a burst shape, the interval where the transmitted signal is no-signal is correlated. The disadvantage of the present invention is that it is possible to obtain accurate correlation values by stopping the operation of the device.

(Example) Next, the present invention will be explained in detail with reference to the drawings.

FIG. 1 is a block diagram showing one embodiment of the present invention.

Here, it is assumed that the circuits shown in FIG. 1 are connected oppositely via a two-wire transmission line 17. If the target is subscriber Google, one will be installed on the central office side and the other on the subscriber side. Here, to simplify the explanation, baseband transmission will be assumed and FIG. 1 will be explained as a subscriber side circuit. In FIG. 1, a binary data sequence is supplied to input terminal l, and input buffer 2 The output of the input buffer 2 is supplied to a transmitter 3 and an adaptive digital filter (ADF) 6.

The binary data sequence is converted into a transmission line code in the transmitter 3 and sent to the two-wire transmission line 17 via the hybrid transformer 4.The output of the input buffer 2 and the output of the multiplier 13 are sent to the zero adaptive digital filter 6. This adaptive digital filter 6 is supplied with an output D/A converter 7 and an output D/A converter 7 of the adaptive digital filter 6, which generates an echo replica corresponding to the echo caused by the impedance mismatch of the hybrid transformer 4. The signal is supplied to the code extraction circuit 10.

The echo replica output from the adaptive digital filter 6 is converted into an analog signal by the D/A converter and supplied as one input to the subtracter 8.

On the other hand, the echo component of the transmission signal generated by the transmitter 3 is passed through the hybrid transformer 4 to the low-pass filter 5.
is supplied to Furthermore, the other party Bl opposite to the circuit in Fig. 1
The received signal sent out from I is supplied to a low-pass filter 5 via a two-wire transmission line 17 and a hybrid transformer 4. The signal whose frequency components outside the desired band are suppressed by the low-pass filter 5 is supplied to one input of the subtracter 8 . The subtracter output is supplied to the receiving section 14 and code detector 9. The sign detector 9 detects only the sign of the input signal, and the sign is supplied to the correlator 11 and the multiplier 13.

The correlator 11 includes a code detection circuit 9. The output signals of the code extraction circuit 1o and the control unit 16 are supplied, and the correlation value between the output signal of the code extraction circuit 1o, which is the code of the echo replica, and the output signal of the code detection circuit 9, which is the code of the error signal. Calculate. The operation of the correlator 11 is determined by a timing signal supplied from the control section, and the on/off operation of the 0 correlator 11, which periodically repeats the on/off operation, will be described in detail later. The output signal of the correlator 11 is sent to the multiplier 1
2, the output of the multiplier 13 is multiplied by 2α (α is a positive constant) and then supplied to the multiplier 13 as the tap correction amount of the adaptive digital filter 6. Supplied.

On the other hand, the error signal which is the output of the subtraction circuit 8 is supplied to the receiving section 14 and is inversely converted from the transmission path code into a binary data series. Further, the receiving section 14 performs clock extraction, data separation, burst detection, etc., and supplies the results to the control section 16. Furthermore, the burst-like binary data series, which is the output signal of the receiving section 14, is supplied to an output buffer, and is inversely converted into a non-burst binary data series, which appears at the output terminal 18.

Next, we will explain the operation of each part in Figure 1 in Figure 2 (Tun (b)
A more detailed explanation will be given with reference to J and FIGS. 3(at to el). Figure 2 (al is a code called WAL-1 or pi-phase, and binary data '
1 for “O” and @1m, respectively, as shown in the figure.
A perfectly balanced DC pulse waveform is assigned within a time width T corresponding to bits. Figure 2 (bl is a code called WAL-2 code, which is completely DC-balanced within T like the WAL-1 code, and has a slightly lower signal spectrum than the WAL-1 code. 2 using Loecho canceller technology, which has the characteristic of
A binary code as shown in Fig. 2 (aly (b)K) is often used as a transmission line code when realizing line bidirectional digital transmission.The reason is that the number of tap coefficients of an adaptive filter is This is because the better the DC balance of the transmission line code is, the less DC is needed, and therefore the hardware scale can be reduced.Now, we will use a binary code as the transmission line code and calculate the convergence of the echo canceller using the sine algorithm. If this is done, the problem arises that the adaptive operation of the echo canceller becomes impossible due to the presence of the received signal.To solve this problem, conventional methods have used the error signal (
When performing 1-pit Ni quantization on the signal (=echo echo replica + received signal), a method is known in which random noise of the same bell frequency as the received signal is added and then 1-pit quantization is performed. In order to avoid this problem, the present invention adopts a method in which transmitting signals are sent in bursts, and the transmitting and receiving circuits of the opposing devices are configured so that the silent periods of the transmitted and received signals do not overlap in time. are doing. Next, a method of sending out a transmission signal will be explained.

FIG. 3 shows a timing chart of transmitted and received signals. The figure (,) is a transmission binary data series,
F indicates the frame length (0, for example, the bit rate of a binary data series) f: fobit/S, If a frame is assembled in units of n bits, then F = n/fosec. FIG. 3(b) shows the transmitted signal,
An example of the WAL-1 code shown in FIG. 2(a) is shown in FIG. Assuming that the time width of the no-signal period is τ, the bit rate of the transmitted signal is f0/(1-τf0/n) bit/s
It increases by 1/(l-τf0/n) times.

The binary data series in Figure 3(a) is the large sword terminal I in Figure 1.
K is supplied. In the input buffer 2 of FIG.
According to the control signal, the bit rate of the input signal is increased by a factor of 17 (l-τfo/n) in frame units, and the signal is output in a form that allows the no-signal period to be distinguished. Therefore, the output signal from the output buffer 2 is expressed as a ternary or binary 2-pit signal, and is supplied to the transmitter 3 and the adaptive digital filter 6. The output signal of the transmitter 3 is the third
The operation of the correlator in FIG. 3(c) will be described later.
First, we will explain ◎ Fig. 3(d) shows a timing chart of the received signal.Synchronization is performed so that the no-signal periods of each of the transmitted and received signals do not overlap.

Since the received signal is in a no-signal state in the no-signal period shown in Figure 3(d), the adaptive
Upon convergence of the filter, its adaptive behavior will be guaranteed. Figure 3 (el) shows the received binary data series, in which the data in the signal section of the received signal is identified, the bit rate is lowered by (1-τfo/n) times, and the original bit Figure 3 (e
The received binary data series of 1 appears at the output terminal 15 in FIG. 1. In the output buffer 15 of FIG. Returned to value data series. Note that the input buffer 2 and the output buffer 3 can be easily realized by using, for example, a digital element called first-in first-out. Next, the operation of the echo canceller will be explained in detail.In FIG. 1, the adaptive digital elements 'D/A converter 7, subtractor 8. A closed loop circuit composed of the sign detection circuit 9 and the multiplier 13 operates adaptively to reduce the echo component contained in the output signal of the subtracter 8.
This adaptive operation is realized by updating the tap coefficients of the adaptive digital film 6.

- The magnitude of the correction amount for each tap coefficient in the multiplier 1
This corresponds to the absolute value of the output signal of No. 3. In FIG. 1, the code extraction circuit 10. If the correlator 11 and the multiplier 12 are deleted and the input of the multiplier 13 supplied from the multiplier 12 is shifted from 2α, the output signal of the subtracter 8, ie, This is because when feeding back the error signal, only the code information of the error signal is fed back to the code detection circuit 9. The present invention differs from the conventional configuration in that, as shown in FIG. will be fed back to the adaptive digital filter 6. Therefore, the output value of the correlator 11 will be a large value when the residual echo is large, and a small value when the residual echo is small. Can be shortened. Here, in addition to the two signals to be correlated, namely the output signal of the code detection circuit 9 and the output signal of the code extraction circuit 10, a control signal is inputted to the correlator 11 of FIG. There is.

A timing chart of control signals input from the control section 16 is shown in FIG. 3(d). In response to this control signal, the correlator 11 is configured to operate only during a time period in which a transmission signal is present within one frame, and to stop its operation during a time period when there is no transmission signal. Such an on/off operation of the correlator 11 means that the error signal does not include an echo because there is no echo itself during the time interval when the transmitted signal is silent, and the error signal sign and echo replica are not included. This is because there is no point in taking a correlation with the sign of . Therefore, during the time period in which there is no transmission signal, the output signal of the correlator 11 holds the value immediately before the correlator 11 turned off. An accurate correlation value can be obtained by such on/off operation of the correlator 11.

In the above description, an example was given in which a binary code is used as a transmission line code, but it is clear that the present invention can also be applied to a ternary code such as a bipolar code or a multi-value code. In this case as well, the transmitted signal is converted into a burst,
It is sufficient to stop the operation of the correlator in the no-signal period. In FIG. 1, this can be easily realized by modifying the configurations of the transmitter 3 and the receiver 14 in accordance with the adopted transmission line code.

(Effects of the Invention) As described in detail above, according to the present invention, by converting the transmitted signal into a burst signal at a constant frame period, the no-signal period is left alive, and the same conversion is performed. The adaptive operation of the echo canceller can be guaranteed by using the no-signal period of the received signal. Therefore, since the circuit that adds random noise, which is necessary when implementing the conventional sign algorithm, becomes unstable, it is possible to reduce the hardware scale. In addition, by adaptively changing the magnitude of the tap correction amount of the adaptive filter using the correlation value between the sign of the error signal and the sign of the echo replica, υ achieves significant convergence compared to conventional sine algorithms. It becomes possible to realize time reduction. At this time, an accurate correlation value can be obtained by stopping the operation of the correlator during a time period in which there is no transmission signal, thereby making it possible to further shorten the convergence time. .

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing one embodiment of the present invention, and FIG. 2 (
al, (Example of pulse waveform of BL digit transmission line code, 3rd
Figures (aJ to (el) indicate the timing chart in Figure 1. In the figure, 1 input terminal, 2 input buffer, 3 transmitter, 4 hybrid circuit, 5 low-pass filter, 6 is an adaptive digital filter, 7 is a D/A
Converter, 8 is a subtracter, 9 is a code conversion circuit, 10 is a code extraction circuit, 11-digit correlator, 12-digit and 13-digit multiplier, 1
4 is a receiving section, 15 is an output buffer, curtain 6 is a control section, 1
7 indicates the 2-wire transmission line, 18 indicates the output terminal, and O%
% 71 (a) O'1 111 La (b)

Claims (1)

[Claims] This is an echo removal method that removes echoes leaking from the transmitting side to the receiving side on the 4-wire side of a 2-wire/4-wire conversion circuit, and generates an echo replica by adaptive filtering processing based on the transmitted signal and the received signal, In the echo removal method of removing the echo by subtracting the echo replica from a signal in which the received signal and the echo are mixed (mixed signal), the mixed signal and the echo replica signal are removed during a period when the transmitted signal is not a signal. The correlation between the sign of the difference signal of An echo cancellation method characterized in that the modification of the coefficients is stopped.