JPH07147555A - Automatic equalizer - Google Patents

Abstract

PURPOSE:To obtain the automatic equalizer matching with the transmission characteristic and tracing with it. CONSTITUTION:A peak value A of an equalization signal is calculated by an output of an equalizer 4 correcting deteriorated components of input signals {yk} converted by an A/D converter 2. An error signal component {rk} is extracted from the result discriminated by a tentative discrimination device 9 based on the equalization signal and the peak level A. The error signal components {rk} and the input signals {yk} are given to a correlation coefficient calculation device 11 and the sum of squares of the input signals {yk} is calculated by a sum of squares calculation device 15, and a correction term {ej} of the equalization coefficient is calculated from the result. The correction term is used to revise an equalization coefficient {cj} at the present time into a new equalization coefficient {cj'} sequentially. Thus, a change in the characteristic of a transmission line such as a head tape or the like is traced and more accurate equalization suitable for each characteristic is attained and a decoding error is reduced.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an automatic equalizer for effectively equalizing and restoring a deteriorated signal component when reproducing or receiving a digitally recorded or transmitted signal.

[0002]

2. Description of the Related Art Consider the case of recording / reproducing digital data on a magnetic tape. In the case of magnetic recording, since the recording / reproducing system shows differential characteristics, the recorded recording code sequence is {b _k }
Then, the ideal reproduced output sequence {x _k } on the assumption that there is no waveform interference is given by the equation (1).

[0003]

[Number 1] x _k = b _k -b _k-1 However, symbols including a subscript k is the sample value at time _{(k · T R), T} R represents a 1-bit length. In reality, the reproduced waveform has a blunted shape due to the high-frequency component attenuation characteristic in the magnetic recording system. Here, the isolated playback waveform is
Assuming that (t) and h _i = h (i · T _R ), {y _k } of the actual reception sequence is expressed by the equation (2).

[0004]

[Equation 2]

However, n _k 'is a sample value of noise. Rewriting (Equation 2) with frequency components gives the equation (Equation 3).

[0006]

## EQU3 ## Y (f) = X (f) .H (f) .N '(f) Note that X (f), H (f), N' (f) and Y (f) are respectively x It is obtained by Fourier transform of (t), h (t), n '(t) and y (t).

In high-density recording required for a digital VTR, decoding based on the amplitude level of a reproduced signal is necessary, and for this purpose, waveform interference must be removed. As can be seen from (Equation 3), the equalized output Z (f) shown in (Equation 4) is obtained by performing waveform equalization having a characteristic of 1 / H (f) that corrects the high frequency component attenuated in the reproduction process. And the waveform interference can be eliminated.

[0008]

## EQU00004 ## Z (f) = X (f) + N '(f) / H (f) = X (f) + N (f)

[0009]

## EQU5 ## N (f) = N '(f) / H (f), where N (f) is equalization noise. Thus, 1 / H
By using the waveform equalizer having the characteristics of (f),
Waveform interference is eliminated.

However, in such a method, it is necessary to know the characteristics of the head tape in advance. Also, when different tapes are used, their equalization characteristics cannot be changed dynamically. In addition, it is not possible to deal with a change in the characteristics due to the partial deterioration of the characteristics of each tape or the wear of the head. Therefore, an automatic equalizer capable of changing the equalization characteristic has been devised. For example, the method is described in the following document ("Automatic
ic Equalization for Digital Communication ”, Bell
System Technical Journal, p. 547, April 1965).

According to these methods, an ideal equalized waveform predicted from the received signal is obtained, a correlation between the ideal equalized waveform and the received signal is obtained, and an equalization coefficient is set according to the magnitude of this correlation. It is a thing.

[0012]

However, in performing the above method, complicated correlation calculation is required,
Digital signal processing is indispensable in order to ensure accuracy. If digital calculation is performed in such a correlation calculation, there is a problem that high-speed calculation that is one digit or more faster than the data transmission rate is difficult to realize.

The present invention solves the above-mentioned conventional problems, and can automatically change the equalization coefficient and can realize the automatic equalization device at an operation speed as high as the data transmission speed. The purpose is to provide.

[0014]

In order to achieve this object, an A / D converter for converting an input signal into a digital signal, an equalizing means for correcting a deterioration component of the input signal, and the equalizing means are used. From the peak value calculation means for calculating the peak value of the output equalized signal, the judgment means for tentatively judging the equalized signal based on the calculated peak value, and the judgment result and the equalized signal. Error signal component extracting means for extracting the error signal component, correlation coefficient calculating means for calculating a cross-correlation coefficient between the error signal component and the input signal, sum of squares calculating means for calculating the sum of squares of the input signals, A correction term calculation means for calculating a correction term of the equalization coefficient of the equalization means from the calculation results of the correlation coefficient calculation means and the sum of squares calculation means, and an equalization for sequentially updating the equalization coefficient of the equalization means from this correction term. Output the output of the coefficient updating means and the equalizing means Based on the equalization coefficient of the first equalization means and the second equalization means after the first equalization means for correcting the deterioration component of the input signal. Second equalization coefficient updating means for updating the equalization coefficient of the second equalization means and equalization signal output means for outputting the output of the second equalization means are provided.

[0015]

According to the above construction, the equalization coefficient can be adaptively changed to perform efficient automatic equalization, and local changes in the input signal can be dealt with.

[0016]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS A case where digital data is recorded and reproduced on a magnetic tape will be considered here. First, the principle of the present invention will be described. Let {y _k } be the reproduced signal sequence, N be the number of equalization taps, and {c _i } (i = 1, 2, ..., N) be the equalization coefficient, and so on at time t = k · T. {Eq _k } of the rectifier output is
It is expressed by the equation (6).

[0017]

[Equation 6]

When it is considered that the equalization coefficient {c _i } defined by the above equation includes an error, the error is {e _i } (i = 1, 2,
, N), the optimum equalization coefficient {c _i ′} (i = 1, 2, ..., N) that makes the residual waveform interference zero can be defined by the equation (7).

[0019]

By substituting c _i = c _i ′ + e _i (Equation 7) into (Equation 6),

[0020]

[Equation 8]

[0021] Further, if {x _k } is a true signal component and {n _k } is a noise component in {y _k } of the reproduced signal sequence,

[0022]

## EQU9 ## y _k = x _k + n _k . By substituting (Equation 9) into the first term of the final expression of (Equation 8),

[0023]

[Equation 10]

It becomes Since the first term of the final expression of (Equation 10) is a convolutional sum of the optimum equalization coefficient {c _i ′} and the noise-free signal component {x _k }, an ideal equality with no residual waveform interference is obtained. It can be regarded as an output. Therefore, this first term is 0
Alternatively, only three values of ± A are taken. However, A is the peak value (0-P) of the equalized signal. Also, if the second and subsequent terms of the final expression of (Equation 10) are {r _k },

[0025]

[Equation 11]

[0026]

Here, {r _k } for M samples
And {y _k } of the cross-correlation _coefficient {m _j } (i = 1, 2, ...,
N) is calculated.

[0028]

[Equation 12]

In the final expression of (Equation 12), assuming that {n _k } are independent of each other, {n _k } and {x _k } are independent, and {y _k } are independent of each other,

[0030]

[Equation 13]

It becomes In (Equation 13),

[0032]

Since n _kpj ² << y _kpj ² ,

[0033]

[Equation 15]

It becomes here,

[0035]

[Equation 16]

Putting it aside,

[0037]

## EQU17 ## e _j = m _j / s _j .

The above is summarized as follows. The value obtained by temporarily determining the equalized output to 0 or ± A is subtracted from the equalized output.
The correction term is determined by taking the ratio of the cross-correlation coefficient between the result and the input signal and the sum of squares of the input. A new equalization coefficient is updated by subtracting this correction term from the current equalization coefficient. By repeating this, equalization using a new equalization coefficient can always be performed.

Next, embodiments of the present invention will be described in detail with reference to the drawings. FIG. 1 shows a block diagram of a first embodiment of the present invention. In FIG. 1, 1 is an input signal terminal, 2 is an A / D converter, 3 is a shift register, 4 is an equalizer, 5 is an equalization coefficient memory of the equalizer 4, and 6 is an equalizer 4.
, 7 is an adder of the equalizer 4, 8 is a peak value calculator,
9 is a tentative judger, 10 is a subtractor, 11 is a correlation coefficient calculator, 12 is a multiplier of the correlation coefficient calculator 11, 13 is a buffer of the correlation coefficient calculator 11, 14 is a divider, and 15 is a square. Sum calculator, 16 is a squarer,
Reference numeral 17 is a buffer of the square sum calculator 15, 18 is a subtracter, and 19 is an equalization signal output terminal.

The operation of this embodiment will be described below. The input analog signal input to the input signal terminal 1 is A
It is converted into a digital signal by the / D converter 2. Later,
This digital signal is treated as the input signal {y _k }. The input signal {y _k } is input to the shift register 3. The shift register 3 has a capacity of (M + N) words. However, M is the update period of the equalization coefficient, and N is the number of equalization taps. In the shift register 3, when a new signal is input, the oldest information is discarded, each other information is shifted by 1 bit, and the input signal becomes the newest information.

The equalizer 4 comprises an equalization coefficient memory 5 for holding N equalization coefficients, N multipliers 6 and an adder 7, and the equalization coefficient memory 5 is updated at a previous time. Equalization coefficient {c _i } (i = 1, 2, ..., N) (N is the number of equalization taps)
Holding The contents of the latest N words (y _k-1 , y _k-2 , ..., Y _kN ) of the shift register 3 are input to the equalizer 4, which are stored in the respective inputs and the equalization coefficient memory 5. The output {eq _k } of the equalizer 4 is output according to (Equation 6) from the equalization coefficient {c _i } that is present, via the multiplier 6 and the adder 7.

The peak value calculator 8 outputs the output {e of the equalizer 4
q _k } is input. Then, a new peak value is calculated from the output {eq _k } of the equalizer 4 which is the current input of the peak value calculator 8 and the immediately preceding peak value. Let the calculated peak value be A (0-P). The output of the equalizer 4 {e
q _k } and the peak value A that is the output of the peak value calculator 8 as a reference value are input, and one of the three values of 0 or ± A is output as the determination result {d _k } of the provisional determiner 9. To be done.
Specifically, when eq _k <−A / 2, d _k = −A, −A /
When 2 <eq _k <A / 2, d _k = 0, and when A / 2 <eq _k , d _k = A. The subtractor 10 outputs the output of the equalizer 4 {e
q _k } and the determination result {d _k } of the output of the provisional determiner 9 are input, and the latter is subtracted from the former. The result of the subtraction is the error signal component {r _k }.

The correlation coefficient calculator 11 includes N multipliers 12 and N
Consists of a cumulative addition buffer 13 and a shift register 3
To the input signal {y _k }, and the subtractor 10 receives the error signal component {r _k }. Now, in the initial state, it is assumed that the update cycle of the equalization coefficient is switched and the contents of the buffer 13 are cleared to zero. When the error signal component {r _k } which is the output of the subtractor 10 is input to the correlation coefficient calculator 11, the error signal component {r _k } and the j-th information y _{kj of the} shift register 3 are multipliers. It is multiplied by 12, and the multiplication result is added to the contents of the buffer 13. This is performed M times in total each time the error signal component {r _k } is input. Since the contents of the shift register 3 are shifted every time, the correlation coefficient calculator represented by the first equation of (Equation 12) is stored in the buffer 13.
Eleven outputs {m _j } are obtained. Since the output {m _j } of the correlation coefficient calculator 11 for 1 ≦ j ≦ N must be obtained,
The above operation is performed for each j. Also, M
When the output {m _j } of the correlation coefficient calculator 11 is output to the divider 14 after the addition for the second time is completed, the contents of the buffer 13 are cleared to zero.

The sum of squares calculator 15 comprises N squarers 16 and N cumulative addition buffers 17, and the input signal {y _k } is input from the shift register 3. Now, in the initial state, it is assumed that the content of the buffer 17 has been cleared to zero. According to the system clock, the shift register 3
The j-th information y _kj is squared by the _squarer 16, and the calculation result is added to the contents of the buffer 17. Since the contents of the shift register 3 are shifted once, the output of the square sum calculator 15 which is the sum of squares of M samples defined by (Equation 16) in the buffer 17 is {s _j } ( j = 1, 2, ...
..., N) is calculated. In this case as well, the output {s _j } of the square sum calculator 15 for 1 ≦ j ≦ N must be obtained, and therefore the same operation is performed for each j. After the M-th addition, the output {s _j } of the square sum calculator 15 is the divider 1
When it is output to 4, the contents of the buffer 17 are cleared to zero.

[0045] to the divider 14, the output of square sum calculator 15 and the output {m _j} of the correlation coefficient calculator 11 {s _j} is input,
The correction term {e _j } (j = 1, 2, ..., N) of the equalization coefficient obtained by dividing the former by the latter is output for each j. Subtractor 18
The equalization coefficient {c _j } at the present time and the correction term {e _j } of the equalization coefficient which is the output of the divider 14 are input to, and the latter is subtracted from the former according to (Equation 7), and the result is Let a new equalization coefficient {c _j ′}.

By repeating the above operation every time the input of M times is completed, the automatic equalization can be adaptively performed. The output {eq _k } of the equalizer 4 is the output of the entire equalizer and is output from the equalized signal output terminal 19.

In the first embodiment described so far, the equalization coefficient is updated every update cycle M of the equalization coefficient, and the updated equalization coefficient is used for the next equalization. When the reproduction output is stable and its amplitude does not change so much, there is no practical problem in the first embodiment. However, it is possible that the amplitude change of the reproduction signal is large due to track bending during recording, tracking failure during reproduction, or damage to the tape. In such a case, the automatic equalizer of the first embodiment may not be able to calculate the correct equalization coefficient,
There is room for further improvement.

Therefore, as a second embodiment, an automatic equalizer capable of coping with the above problems will be described below. FIG. 2 shows a block diagram of the second embodiment of the present invention. Further, the same effects as those in FIG. 1 are designated by the same reference numerals, and detailed description thereof will be omitted.
In FIG. 2, 20 is a second equalizer and 21 is a second equalizer 20.
Of the second equalizer 20, 22 is a multiplier of the second equalizer 20, and 23 is an adder of the second equalizer 20.

The operation of the second embodiment will be described below. In the second embodiment, the output of the equalizer 4 is not used as the output of the equalizer. The input signal {y
_k } is equalized and output from the equalized signal output terminal 19.
The equalization coefficient of the second equalizer 20 is determined as follows. A certain period M '(M'>> M, defined by the system,
The equalization coefficient held in the equalization coefficient memory 5 is copied to the second equalization coefficient memory 21 at every update interval of the equalization coefficient of the equalizer 4). Then, the second equalizer 20 performs equalization with the equalization coefficient until the next equalization coefficient is updated.

According to this method, even if the equalization coefficient of the equalizer 4 which is sequentially updated becomes a wrong value temporarily, the second equalizer 20 is not affected by it and the first equalizer directly connected to the output. The second equalizer 20 can perform correct equalization.

In this embodiment, the input signal {y _k } is A / D converted and the digital signal processing is performed thereafter, but it is also possible to handle the analog signal as it is.
Further, when digital signal processing is performed, the equalized result is output as a digital signal, so that it is not necessary to newly perform A / D conversion when performing Viterbi decoding. That is,
One A / D converter can perform digital equalization and digital decoding.

[0052]

According to the present invention, it is possible to obtain an automatic equalizer which can perform equalization adapted to the characteristics of a transmission line such as a head tape and can follow changes in the characteristics. Therefore, more accurate equalization can be performed and decoding error can be reduced.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration in a first exemplary embodiment of the present invention.

FIG. 2 is a block diagram showing a configuration in a second exemplary embodiment of the present invention.

[Explanation of symbols]

1 ... Input signal terminal, 2 ... A / D converter, 3 ... Shift register, 4 ... Equalizer, 5 ... Equalization coefficient memory,
6, 12 ... Multiplier, 7 ... Adder, 8 ... Peak value calculator,
9 ... Tentative judger, 10, 18 ... Subtractor, 11 ... Correlation coefficient calculator, 13, 17 ... Buffer, 14 ... Divider, 15 ... Square sum calculator, 16 ... Squarer, 19 ... Equalized signal output Terminal,
20 ... Second equalizer, 21 ... Second equalization coefficient memory, 22
… Second multiplier, 23… Second adder.

Claims (2)

[Claims] 1. A for converting an input signal into a digital signal
A / D converter, an equalizing means for correcting the deterioration component of the input signal digitized by the A / D converter, and a peak value (denoted as A) of the equalized signal output by the equalizing means. A peak value calculating means for calculating, a judging means for tentatively judging the equalized signal to 0 or ± A based on the peak value A calculated by the peak value calculating means, and a judgment result by the judging means. An error signal component extraction unit that extracts an error signal component from the equalized signal, a correlation coefficient calculation unit that calculates a cross-correlation coefficient between the error signal component and the input signal, and a sum of squares of the input signal A sum of squares calculating means, a correction term calculating means for calculating a correction term for the equalization coefficient of the equalizing means from the calculation results of the correlation coefficient calculating means and the sum of squares calculating means, and the correction term Equalization coefficient update that sequentially updates the equalization coefficient of the equalization means And an equalization signal output means for outputting the output signal of the equalization means. 2. A for converting an input signal into a digital signal
And an A / D converter, a first equalizing means and a second equalizing means for correcting the deterioration component of the input signal digitized by the A / D converter, and the first equalizing means. Based on the peak value A calculated by the peak value calculating means and the peak value A calculated by the peak value calculating means. Determination means for determining, error signal component extraction means for extracting an error signal component from the determination result by the determination means and the equalized signal, and a phase for calculating a cross-correlation coefficient between the error signal component and the input signal Relation number calculation means, square sum calculation means for calculating the sum of squares of the input signals, correction coefficient of the first equalization means from the calculation results of the correlation coefficient calculation means and the sum of squares calculation means Correction term calculation means for calculating the term,
First equalization coefficient updating means for sequentially updating the equalization coefficient of the first equalization means from the correction term, and the second equalization based on the equalization coefficient of the first equalization means An automatic equalization device comprising: a second equalization coefficient updating means for updating the equalization coefficient of the means; and an equalization signal output means for outputting an output signal of the second equalization means.