JP2000224080A - Device and method for equalizing line - Google Patents

Abstract

PROBLEM TO BE SOLVED: To obtain a good line equalizing characteristic, regardless of line characteristics. SOLUTION: Based on a received signal 101 delayed by means of a delaying device 103 and an identified error 102, a filter factor updating section 106 updates the tap factor of a tap by using an LMS algorithm, and in addition, a factor control section 105 having a prescribed threshold changes the tap factor generated by means of the updating section 106 to the same level as that of the threshold.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

TECHNICAL FIELD The present invention relates to a TCM (Time Com
In a digital subscriber line transmission system (hereinafter, referred to as a TCM transmission system) of the Compressed Multiplex system,
The present invention relates to a line equalizer suitable for compensating for distortion caused by a transmission line and a method of equalizing the same.

[0002]

2. Description of the Related Art As shown in FIG. 7, data transmission in a TCM transmission system is such that communication between a station and a subscriber is performed in a time-division manner on a single transmission line.

[0003] Such a TCM transmission system is provided with a received signal processing section having a line equalizer to which a time for training is allocated.

A training signal used for training is transmitted in synchronization with 320 kHz. The training signal has a fixed pattern in which the header has 16 bits and takes +1 and -1 every 8 bits. For this reason, the center frequency of the transmission signal is 20 kHz.

[0005] A training signal at the time of training is transmitted from an opposite transmitting unit. During normal communication,
A random signal is transmitted from the opposite transmitting unit. Its center frequency is 160 kHz. AMI (Al
ternate Mark Inversion) is used.

FIG. 8 shows an example of such a line equalizer.

The line equalizer 900 shown in FIG. 8 is composed of an N-tap (N is a natural number) transversal filter. Updating of tap coefficients is generally performed by an LMS algorithm.

The line equalizer 900 includes a delay unit 903, a multiplier 904, a filter coefficient update unit 906, and an adder 907.
It has.

[0009] Received signal 101 is applied to multiplier 904 and filter coefficient updating section 906 via delay unit 903. The identification error 102 is provided to the filter coefficient update unit 906.

The signal updated by filter coefficient updating section 906 is provided to multiplier 904. Multiplier 904
Received signal 101 given via delay unit 903 at
And the signal updated by the filter coefficient update unit 906 are multiplied. The signals multiplied by each multiplier 904 are added by an adder 907 and output as a code identification signal 108.

[0011]

The above-described training of the line equalizer 900 is performed on a training signal having a center frequency of 20 kHz as described above. In this case, depending on the line characteristics, the precursor may not be shaped as shown in FIG. 9A.

This is because the training signal has a fixed pattern, and the portions A and B in FIG. 9 cancel each other out, and the reception level at the sample point becomes zero. Focusing on the tap coefficient of the line equalizer 900 at this time, the leading tap coefficient has a positive value (when the center tap is set to 4) as shown in FIG. 10, which causes an increase in noise. I was

As described above, in the training of the line equalizer 900, since the noise increases, there is a problem that the line equalization characteristics deteriorate due to the line characteristics.

SUMMARY OF THE INVENTION The present invention has been made in view of such circumstances, and provides a line equalizer capable of obtaining good line equalization characteristics irrespective of line characteristics and a method of equalizing the same. To make it possible.

[0015]

According to a first aspect of the present invention, there is provided a line equalizer including a transversal filter having N taps (N is a natural number), and adapted to adaptively equalize a waveform of a received signal. An equalizer, cascaded to each other,
A plurality of delay units for giving a predetermined delay to the received signal;
Based on the delayed received signal and the identification error, a filter coefficient update unit that updates the tap coefficient of the tap by the LMS algorithm, and has a predetermined threshold value, and calculates the value of the tap coefficient generated by the filter coefficient update unit. , And a coefficient control unit that changes the value to the same value as the threshold value. When the tap coefficient is the i-th (i is 1 to N) tap coefficient C _i (n) at time n, C _i ′ (n + 1) = C _i (n) + α × e (n) × D
_{The i-} th tap coefficient candidate C _i ′ at time n + 1 calculated by _{i + 1} (n)
(E (n) is the discrimination error at time n, D _{i + 1} (n) is the output of the ( _{i + 1} ) th delay unit, α is the gain, and the threshold is 0. ). Also, the tap coefficient C at time n + 1
_i (n + 1) is determined by the way of limiting tap coefficient candidates. When the expected tap coefficient is negative, C _i (n + 1) = 0 if C _i ′ (n + 1)> 0: (Limit tap coefficient candidates) C _i (n + 1) if C _i ′ (n + 1) ≦ 0 = C _i '(n + 1): (no restriction on tap coefficient candidates). When the expected tap coefficient is positive, if C _i ′ (n + 1) ≧ 0, C _i (n + 1) = C _i ′ (n + 1): (no restriction is imposed on the tap coefficient candidates) If C _i ′ (n + 1) <0 For example, C _i (n + 1) = 0: (limit tap coefficient candidates).
In addition, the line equalizer may include a decision feedback equalizer that suppresses a post-cursor after the determination of a solitary wave. A line equalizer method according to claim 5, wherein the line equalizer includes an N-tap (N is a natural number) transversal filter and adaptively equalizes the waveform of a received signal. A first step of cascading each other and providing a predetermined delay to a received signal;
A second step of updating the tap coefficient of the tap by the S algorithm, and a third step of changing the value of the tap coefficient generated by the update to the same value as the threshold value, having a predetermined threshold value And characterized in that: In the third step, the tap coefficient is set to the i-th (i is 1
When the tap coefficients to N) was _{C i (n), C i} '(n + 1) = C i (n) + α × e (n) × D
_{The i-} th tap coefficient candidate C _i ′ at time n + 1 calculated by _{i + 1} (n)
(E (n) is the discrimination error at time n, D _{i + 1} (n) is the output of the ( _{i + 1} ) th delay unit, α is the gain, and the threshold is 0. ) A fourth step may be included. Further, in the fourth step, the tap coefficient C _i (n + 1) at time n + 1 is determined by adding a restriction to the tap coefficient candidate. When the expected tap coefficient is negative, C _i (n + 1) = 0 if C _i ′ (n + 1)> 0: (Limit tap coefficient candidates) C _i (n + 1) if C _i ′ (n + 1) ≦ 0 = C _i '(n + 1): (no restriction on tap coefficient candidates). When the expected tap coefficient is positive, if C _i ′ (n + 1) ≧ 0, C _i (n + 1) = C _i ′ (n + 1): (no restriction is imposed on the tap coefficient candidates) If C _i ′ (n + 1) <0 For example, it is possible to include a step of calculating by two methods: C _i (n + 1) = 0: (limit tap coefficient candidates). Further, the first to third steps may include a step of suppressing the post-cursor after the determination of the solitary wave. In the line equalizer and the equalization method according to the present invention, based on the received signal delayed by the delay unit and the identification error, the filter coefficient update unit updates the tap coefficient of the tap by the LMS algorithm, The value of the tap coefficient generated by the filter coefficient updating unit is changed to the same value as the threshold value by the coefficient control unit having the threshold value.

[0016]

Embodiments of the present invention will be described below.

(First Embodiment) FIG. 1 is a block diagram showing a line equalizer according to a first embodiment of the present invention, and FIG.
3 is a block diagram showing a received signal processing unit according to the first embodiment of the line equalizer of the present invention, and FIGS. 3 and 4 are diagrams for explaining the operation of the line equalizer of FIG. is there.

The line equalizer 100 shown in FIG.
03, a multiplier 104, a coefficient control unit 105, a filter coefficient update unit 106, and an adder 107.

The delay unit 103 is cascade-connected, and applies a predetermined delay to the received signal 101.
Further, the delay units 103 are composed of N + 1 (N is a natural number). Further, the delay unit 103 supplies the received signal 101 to the multiplier 104 and the filter coefficient updating unit 106.

The multiplier 104 outputs the output of the i-th (i is 1 to N) filter coefficient updating unit 106 and the i-th delay unit 10
3 is an input. The coefficient control unit 105 receives an output of the i-th filter coefficient update unit 106 as an input.

The filter coefficient update unit 106 receives the output of the (i + 1) th delay unit 103, the output of the i-th coefficient control unit 105, and the identification error 102 as inputs. The adder 107
The outputs of the multipliers 1 to N are input.

The line equalizer 100 is composed of a transversal filter, and includes a received signal 101 and an identification error 102
, And adaptively performs waveform equalization on the received signal 101. The tap coefficient of each tap is generated by the filter coefficient update unit 106 and the coefficient control unit 105. The coefficient control unit 105 can change the value of the tap coefficient generated by the filter coefficient update unit 106 to the same value as the threshold.

FIG. 2 shows a reception signal processing section in the TCM transmission system.

The received signal processing unit 201 shown in FIG.
D converter 204, gain adjustment unit 205, line equalizer 10
0 and a code identification unit 208 are provided.

The A / D converter 204 converts the received signal 101 from the transmitting unit 200 obtained via the line 202 into a digital signal.

The gain adjustment unit 205 is provided for the A / D converter 20
4 is given a gain. At this time, the peak level of the signal is set to match the threshold level of the code identification unit 208. For example, if the identification threshold is 0.5, a flat gain is set such that the peak value is 1.

The line equalizer 100 includes a gain adjustment unit 205
And the discrimination error 102 are input. The line equalizer 100 compensates for the line loss and equalizes the solitary wave response of the received waveform so as to have a roll-off characteristic as shown in FIG.

The code identification unit 208 receives the output of the line equalizer 100 as an input. The code identification unit 208 performs code identification and calculates the identification error 102. The identification error e (n + 1) at the time n + 1 is calculated by the equation (2).

E (n + 1) = D _in (n) −2.0 × th (n) × D _out (n) (2)

Here, D _in (n) is the code identification signal 108 at time n, th (n) is the code identification threshold (signal inside the code identification unit 208) at time n, D
_out (n) is the code identification result 210 at time n.

Next, the operation of the line equalizer 100 when the number of taps is 7 will be described.

The line equalizer 100 is composed of a transversal filter as described above. Line equalizer 10
0 inputs the received signal 101 and the identification error 102, and adaptively equalizes the received signal 101.

For example, as an initial value, 1.0 is set for the center tap (fourth tap coefficient), and 0.0 is set for the other taps. The tap coefficient of each tap is generated by the filter coefficient update unit 106 and the coefficient control unit 105.

The filter coefficient update unit 106 updates the tap coefficients by the LMS algorithm (for the LMS algorithm, see “J. McCool and B. Widrow,“ Prin
ciples and applications of adaptive filters: A tut
orial review, "Naval Undersea Center, San Diego, C
A, Tech. Publ. 530, Mar. 1977. ").

The coefficient control section 105 has a threshold value (threshold 0) therein, and changes the value of the tap coefficient generated by the filter coefficient control section 106 to the same value as the threshold value. Can be.

The change of the tap coefficient is performed as follows.

Assuming that the i-th tap coefficient at time n is C _i (n), the i-th tap coefficient candidate C _i ′ (n
+1) is calculated by equation (1).

C _i ′ (n + 1) = C _i (n) + α × e (n) × D _{i + 1} (n) (1)

Here, e (n) is an identification error 1 at time n.
02, D _{i + 1} (n) is the output of the ( _{i + 1} ) th delay device 103, and α is the gain. The threshold is zero.

Tap coefficient C _i (n + 1) at time n + 1
Is calculated by the following two methods, depending on how a restriction is applied to the tap coefficient candidates.

[0041] If the tap coefficients expected negative _{C i '(n + 1)} > 0 if _{C i (n + 1) =} 0 to limit the :( tap coefficient _{candidates) C i' (n + 1} ) ≦ 0 if C _i (n + 1) = C _i '(n + 1): (no restriction on tap coefficient candidates)

[0042] If the expected tap coefficient is positive, C _i ′ (n + 1) ≧ 0, C _i (n + 1) = C _i ′ (n + 1): (no restriction on tap coefficient candidates) If C _i ′ (n + 1) <0 If C _i (n + 1) = 0: (limit tap coefficient candidates)

As described above, by restricting the update process of the tap coefficients, even if the update of the tap coefficients is performed on the unique training pattern, good characteristics can be obtained when equalizing a random signal.

The control of the tap coefficient is performed as follows.

A characteristic of the transmission line is that the higher the frequency, the greater the loss. Therefore, the line equalizer 1
The equalization characteristic of 00 is a high-pass filter characteristic. As shown in FIG. 4, a coefficient that alternately repeats positive and negative values is given to the tap coefficient in the transversal filter. Thereby, a high-pass filter can be realized.

Here, in the TCM digital subscriber transmission system, the training signal is a fixed pattern signal having a center frequency of 20 kHz. When training the line equalizer 100 with a fixed pattern, good equalization characteristics may not be obtained for random signals. By the way, the center frequency of the random signal is 16
0 kHz. The cause of this is the first that should be negative.
Is positive as shown in FIG.

Therefore, only the first tap coefficient is limited by the coefficient control unit 105 (set the threshold value to 0).
When the tap coefficient becomes a positive value as shown in FIG. 10A, by setting the tap coefficient to 0, deterioration of the equalization characteristic can be prevented. However, the other tap coefficients have no restriction.

As described above, in the first embodiment, based on the received signal 101 and the identification error 102 delayed by the delay unit 103, the filter coefficient updating unit 106
The tap control of the tap is performed by an LMS algorithm, and the coefficient control unit 1 having a predetermined threshold value is updated.
05, the value of the tap coefficient generated by the filter coefficient updating unit 106 is changed to the same value as the threshold value, so that good line equalization characteristics can be obtained regardless of the line characteristics.

(Second Embodiment) FIG. 5 is a diagram showing a line equalizer according to a second embodiment of the present invention.

In the second embodiment, as shown in FIG. 5A, when the first tap coefficient is positive, the first tap coefficient is set to 0, and when the second tap coefficient is negative, , And the second tap coefficient is set to 0.

That is, assuming that the center tap is the fourth tap, in a filter having high-pass characteristics, the first tap coefficient is negative and the second tap coefficient is positive, as shown in FIG. Become. Therefore, as described above, the first
Is positive, the first tap coefficient is set to 0,
When the second tap coefficient is negative, good characteristics can be obtained regardless of the line characteristics by controlling the second tap coefficient to be zero.

(Third Embodiment) FIG. 6 is a block diagram showing a received signal processing section according to a third embodiment of the line equalizer of the present invention.

The received signal processing unit 601 shown in FIG.
D converter 604, gain adjuster 605, line equalizer 60
6, a code identification unit 607, a decision feedback equalizer 610, and a subtractor 612.

The A / D converter 604 converts a reception signal 603 obtained from the opposite transmission unit 600 via the line 602 into a digital signal. The gain adjusting unit 605
A gain is given to the output of the / D converter 604.

The line equalizer 606 includes a gain adjuster 605
And the discrimination error 608 are input. Subtractor 61
2 is the output of the line equalizer 606 and the decision feedback equalizer 6
The subtraction process is performed by using the output of FIG.

The code identification section 607 receives the output of the subtractor 612 as an input and outputs a code identification result 609. The decision feedback equalizer 610 outputs the code identification result 609 and the identification error 6
08 as an input, and outputs a decision feedback equalizer output 611.

Here, the line equalizer 606 has the same function as the line equalizer 100 shown in FIGS. For the decision feedback equalizer 610, see (CARLOS A. BELFIORE, JOHN
H.PARK, JR., "Decision Feedback Equalization", PROC
EEDINGS OF THE IEEE, VOL. 67, NO.8, PP.1143-1156, AUG
AST 1979.).

The decision feedback equalizer 610 suppresses the post-cursor after the determination of the solitary wave. The line equalizer 606 shapes the waveform of the precursor before the determination point.

Even in such a configuration, good characteristics can be obtained regardless of the line characteristics.

[0060]

As described above, according to the line equalizer and the equalizing method according to the present invention, the filter coefficient updating unit performs tapping of taps by the LMS algorithm based on the received signal and the identification error delayed by the delay unit. Since the tap coefficient is updated and the value of the tap coefficient generated by the filter coefficient updating unit is changed to the same value as the threshold value by the coefficient control unit having a predetermined threshold value, the line characteristic is changed. Good line equalization characteristics can be obtained regardless of this.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a first embodiment of a line equalizer according to the present invention.

FIG. 2 is a block diagram illustrating a reception signal processing unit according to the first embodiment of the line equalizer of the present invention.

FIG. 3 is a diagram for explaining an operation of the line equalizer of FIG. 1;

FIG. 4 is a diagram for explaining an operation of the line equalizer of FIG. 1;

FIG. 5 is a diagram showing a second embodiment of the line equalizer of the present invention.

FIG. 6 is a block diagram illustrating a reception signal processing unit according to a third embodiment of the line equalizer of the present invention.

FIG. 7 is a diagram for explaining data transfer in a conventional TCM transmission system.

FIG. 8 is a block diagram showing an example of a conventional line equalizer.

9 is a diagram for explaining the operation of the line equalizer of FIG.

FIG. 10 is a diagram for explaining the operation of the line equalizer of FIG. 8;

[Explanation of symbols]

 100,606 Line equalizer 101 Received signal 102,608 Identification error 103 Delay unit 104 Multiplier 105 Coefficient control unit 106 Filter coefficient update unit 107 Adder 108 Code identification signal 200 Opposite transmission unit 201,601 Received signal processing unit 202 Line 204, 604 A / D converter 205, 605 Gain adjustment unit 208, 607 Code identification unit 210, 609 Code identification result 610 Decision feedback type equalizer 611 Decision feedback type equalizer output 612 Subtractor

Claims (8)

[Claims] 1. A line equalizer configured by an N-tap (N is a natural number) transversal filter and adapted to adaptively equalize a waveform of a received signal. And a plurality of delay units for providing a predetermined delay to the LMS based on the delayed received signal and the identification error.
A filter coefficient updating unit that updates a tap coefficient of the tap by an algorithm, having a predetermined threshold value, and setting the value of the tap coefficient generated by the filter coefficient updating unit to the same value as the threshold value A line equalizer characterized by comprising a coefficient control unit for changing. 2. The coefficient controller according to claim 1, wherein the tap coefficient is:
The i-th (i is 1 to N) tap coefficient at time n is C
_i (n), C _i ′ (n + 1) = C _i (n) + α × e (n) × D
_{The i-} th tap coefficient candidate C _i ′ at time n + 1 calculated by _{i + 1} (n)
(Where e (n) is the discrimination error at time n, D _{i + 1} (n) is the output of the ( _{i + 1} ) th delay unit, α is the gain, and the threshold is 0. The line equalizer according to claim 1, wherein: 3. The tap coefficient C _i (n +
1) is based on how to restrict tap coefficient candidates. If the tap coefficients expected negative _{C i '(n + 1)} > 0 if _{C i (n + 1) =} 0 to limit the :( tap coefficient _{candidates) C i' (n + 1} ) ≦ 0 if C _i (n + 1) = C _i '(n + 1): (no restriction on tap coefficient candidates). When the expected tap coefficient is positive, if C _i ′ (n + 1) ≧ 0, C _i (n + 1) = C _i ′ (n + 1): (no restriction is imposed on the tap coefficient candidates) If C _i ′ (n + 1) <0 3. The method according to claim 2, wherein the calculation is performed in two ways: C _i (n + 1) = 0: (limit tap coefficient candidates).
The line equalizer according to 1. 4. The line equalizer according to claim 1, wherein the line equalizer includes a decision feedback equalizer that suppresses a post-cursor after the determination of a solitary wave. 5. An equalization method for a line equalizer configured by an N-tap (N is a natural number) transversal filter and adaptively performing waveform equalization on a received signal, wherein the line equalizer is cascaded to each other, A first step of providing a predetermined delay to the received signal, and an LMS based on the delayed received signal and the identification error.
A second step of updating a tap coefficient of the tap by an algorithm; and a second step of changing a value of the tap coefficient generated by the update to the same value as the threshold, having a predetermined threshold. 3. A method for equalizing a line equalizer, comprising: 6. In the third step, when the tap coefficient of the i-th (i is 1 to N) at time n is C _i (n), C _i ′ (n + 1) = C _i (n) + α × e (n) × D
_{The i-} th tap coefficient candidate C _i ′ at time n + 1 calculated by _{i + 1} (n)
(Where e (n) is the discrimination error at time n, D _{i + 1} (n) is the output of the ( _{i + 1} ) th delay unit, α is the gain, and the threshold is 0. .) 4th
6. The method according to claim 5, further comprising the step of: 7. The method according to claim 7, wherein the tap coefficient C _i (n + 1) at the time n + 1 is obtained by: If the tap coefficients expected negative _{C i '(n + 1)} > 0 if _{C i (n + 1) =} 0 to limit the :( tap coefficient _{candidates) C i' (n + 1} ) ≦ 0 if C _i (n + 1) = C _i '(n + 1): (no restriction on tap coefficient candidates). When the expected tap coefficient is positive, if C _i ′ (n + 1) ≧ 0, C _i (n + 1) = C _i ′ (n + 1): (no restriction is imposed on the tap coefficient candidates) If C _i ′ (n + 1) <0 7. The equalization method for a line equalizer according to claim 6, further comprising a step of calculating by two methods: C _i (n + 1) = 0: (limit tap coefficient candidates). 8. The equalization of a line equalizer according to claim 5, wherein the first to third steps include a step of suppressing the post-cursor after the determination of the solitary wave. Method.