JP3395283B2 - Adaptive waveform equalizer and adaptive waveform equalization method - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an adaptive waveform equalizing apparatus and an adaptive waveform equalizing method for digitally converting an analog signal transmitted through a transmission line or reproduced from a recording medium and equalizing and outputting the waveform. Regarding

[0002]

2. Description of the Related Art An analog signal transmitted via a transmission line or an analog signal reproduced after being FM (Frequency Modulation) recorded on a recording medium is recorded on the analog signal before being transmitted on the transmission line or on a recording medium. It is different from the analog signal before it was done. Specifically, for example, since the recording medium has a differential characteristic, when the analog signal recorded on the recording medium is reproduced, a differential waveform of the recorded analog signal is obtained.

Further, after the step signal is recorded on the recording medium, the waveform when the recorded step signal is read out is a pulse (reproduction pulse) having a predetermined width. The width of the reproduction pulse determines the frequency of the signal that can be recorded on the recording medium, that is, the amount of information.

[0004]

By the way, for example, the waveforms output when an analog signal recorded on a recording medium is given a signal that changes faster than the above pulse width, the mutual pulses interfere with each other. , The recorded signal cannot be reproduced. In other words, the phase of the 1 or 0 change of the reproduced and equalized digital signal moves due to peak shift or the like in the digital signal when the reproduced analog signal is reproduced and equalized, so that the eye of the eye pattern does not open. When the waveform equalization processing is performed in this state, the number of errors increases when determining 1 or 0, and this waveform equalization filter is not effective.

When reproducing a digital signal recorded on a recording medium, the recorded digital signal is 1
Or, since it has only an amplitude of 0, even if the reproduced pulses interfere with each other to some extent, when the waveform equalization processing is performed, if the waveform equalization filter having a predetermined condition and good performance is used, the recorded digital The signal can be regenerated.

In view of the above situation, the present invention provides an adaptive waveform equalization apparatus and an adaptive waveform equalization method capable of further improving the waveform equalization performance when performing waveform equalization processing. Is.

[0007]

An adaptive waveform equalization apparatus according to the present invention divides signal data transmitted or reproduced from a recording medium into blocks and class-divides the signal data of each class-divided block. A class dividing unit for converting into a numerical value corresponding to the pattern of the signal data, a table reading unit for reading out a filter coefficient from a table of a plurality of pre-stored selectable filter coefficients based on the converted numerical value, The above problem is solved by having a product-sum processing unit that performs a product-sum calculation using the signal data from the class dividing unit and the filter coefficient read from the table reading unit.

Here, analog signal data is used as the signal data transmitted or reproduced from the recording medium, and this analog signal data is converted into digital signal data after being converted by an A / D (analog / digital) conversion circuit. Alternatively, digital signal data may be used as it is as the signal data transmitted or reproduced from the recording medium.

Further, the filter processing section includes a table reading section for reading out a filter coefficient used when binarizing the multivalued digital signal data inputted to the filter processing section from a table stored in advance. And a linear estimation processing unit that binarizes the multi-valued digital signal data by using the read filter coefficient.

Further, the filter coefficients read by the table reading unit are digital signal data before being transmitted or recorded on a recording medium and digital signal data after being transmitted or reproduced from the recording medium. It is characterized in that the multi-value digital signal data before being digitized is divided into blocks, and learning is performed by using these blocked signal data.

In addition, the class dividing unit performs a binarization process on the block processing unit for dividing multi-valued digital signal data into blocks and the block signal data from the block processing unit. A threshold value determination processing unit for determining a threshold value and a binarization for binarizing the blocked signal data from the blocking processing unit using the threshold value from the threshold value determination processing unit. Processing unit and above 2
And a conversion processing unit for converting the binarized signal data for each block from the value conversion processing unit into a corresponding numerical value.

In the adaptive waveform equalization method according to the present invention, the signal data transmitted or reproduced from the recording medium is divided into blocks and class-divided, and the signal data of each class-divided block is converted into a pattern of the signal data. A first step of converting into a corresponding numerical value, a second step of reading out the filter coefficient from a table of a plurality of pre-stored selectable filter coefficients based on the converted numerical value, and the class division And the above signal data,
The above-mentioned problem is solved by having a third step of performing a product-sum operation using the filter coefficient read from the table reading unit.

In the filter processing step, the multi-valued digital signal data input to the filter processing unit is converted into 2 values.
A table reading step of reading a filter coefficient used in digitizing from a table stored in advance,
And a linear estimation processing step of binarizing the multi-valued digital signal data by using the read filter coefficient.

Further, the filter coefficients read in the table reading step are the digital signal data before being transmitted or recorded on the recording medium and the digital signal data after being transmitted or reproduced from the recording medium. It is characterized in that the multi-value digital signal data before being digitized is divided into blocks, and learning is performed by using these blocked signal data.

In addition, the class dividing step is a block processing step for dividing multivalued digital signal data into blocks, and a binarization processing for the blocked signal data from the block processing step. A threshold value determination processing step for determining a threshold value and a binarization for binarizing the blocked signal data from the blocking processing step using the threshold value from the threshold value determination processing step. It is characterized by comprising a processing step and a conversion processing step for converting the binarized signal data for each block from the binarization processing step into a corresponding numerical value.

[0016]

In the present invention, when the signal data transmitted or reproduced from the recording medium is subjected to the waveform equalization processing, the multi-valued digital signal data is divided into blocks and the signal data obtained by class division is converted. The pre-learned and stored filter coefficient is read out using a predetermined numerical value obtained by the above, and the multi-valued digital signal data is subjected to waveform equalization processing by using this filter coefficient and transmitted or recorded. The binarized signal data before being recorded in is obtained.

[0017]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT A preferred embodiment of the present invention will be described below with reference to the drawings.

FIG. 1 shows a schematic configuration of an adaptive waveform equalizer according to the present invention. Also, in this example,
It is assumed that the analog signal data read from the recording medium is input from the input terminal 1 of FIG.

The analog signal data input from the input terminal 1 is input to the A / D (analog / digital) conversion circuit 2. The A / D conversion circuit 2 converts the input analog signal data into multivalued digital signal data by using the recording bit clock rate. The converted multivalued digital signal data is output to the class dividing unit 4 in the waveform equalization processing unit 3.

The class dividing unit 4 divides the input multilevel digital signal data into blocks, and divides the blocked signal data into blocks for each block. Further, the pattern of this class-divided signal data is converted into a predetermined numerical value (class code). The blocked signal data B and the signal P having a predetermined numerical value are supplied to the class-specific adaptive filter processing unit 5.

Here, the filter coefficient determining unit 6 will be described. In the filter coefficient determining unit 6, digital signal data before being recorded on a recording medium and digital signal data after being reproduced from the recording medium, which is multi-valued digital signal data before being binarized. Learning is performed by using and to obtain a filter coefficient, for example, a linear estimation coefficient, which is used when binarizing the multi-valued digital signal data which is the digital signal data reproduced from the recording medium. This linear estimation coefficient is determined by a prediction coefficient estimation device, such as a least squares estimation device. The learning is performed on the signal data obtained by dividing the digital signal data reproduced from the recording medium, that is, the multi-valued digital signal data before being binarized, into blocks. A filter coefficient corresponding to the converted signal data is obtained. The filter coefficient (linear estimation coefficient) determined in this way is a predetermined numerical value when the pattern of the signal data obtained by classifying the blocked signal data corresponding to this filter coefficient is converted, that is, a so-called class code ( Class information) is used as an address and is stored in advance as a table in a ROM (read-only memory).

Therefore, the class-specific adaptive filter processing unit 5 reads out the linear estimation coefficient adapted to the signal data B for each block from the filter coefficient determination unit 6 by using the numerical value of the input signal P. Further, the read linear estimation coefficient is used to perform linear estimation processing on each block of the block signal data, and a binarized signal is output from the output terminal 7.

Next, referring to FIG.
Will be explained. The multivalued digital signal data input from the input terminal 8 is input to the blocking processing unit 9 in the class dividing unit 4. In the blocking processing unit 9,
The digital signal data is divided into blocks, and at this time, the signal data of blocks in the vicinity of the two-dimensional block are converted into one group, that is, converted so that they can be identified. Blocked signal data is signal data B
Is output from the output terminal 14 to the class-specific adaptive filter processing unit 5 of FIG. 1, and is also supplied to the threshold value determination processing unit 10 and the binarization processing unit 11.

In the threshold value determination processing unit 10, the threshold value used when the binarization processing unit 11 performs the binarization processing on the blocked signal data is determined for each block. For example, the average value within the block is used as the threshold value.

In the binarization processing unit 11, the threshold value from the threshold value determination processing unit 10 is manipulated for each block, and using this threshold value, the blocking from the blocking processing unit 9 is performed. The signal data is binarized for each block. This 2
The binarized signal data is output to the packing processing unit 12.

The packing processing unit 12 converts the pattern of the binarized signal data for each block from the binarization processing unit 11 into a predetermined numerical value, a so-called class code (class information). For example, when signal data obtained by dividing n words of signal data of 1 word into 1 block into 1 block is directly divided into classes, 2 ^mn class codes are generated. By performing the value conversion process, the number of class codes is reduced to 2 ⁿ . The predetermined numerical value (class code) is output as a signal P from the output terminal 13.

Next, the class-specific adaptive filter processing section 5 will be described with reference to FIG. The converted predetermined value signal P from the class dividing unit 4 is input from the input terminal 16 to the ROM table reading unit 17. This RO
In the M table reading unit 17, the filter coefficient determining unit 6 reads the linear estimation coefficient in the ROM stored in advance by using the predetermined numerical value (class code). The read linear estimation coefficient is sent to the linear estimation processing unit 18.

The linear estimation processing unit 18 has an input terminal 1
5, the signal data blocked by the blocking processing unit 9 in FIG. 2 is input as the signal data B. By performing the product-sum operation of the signal data blocked and the linear estimation coefficient, Value of digital signal data is 2
The binarized signal data is binarized, and the binarized signal data is output terminal 19.
Will be output.

In this embodiment, the input signal data is used by converting the analog signal data reproduced from the recording medium into digital signal data, but instead of the analog signal data reproduced from this recording medium, It is also possible to input digital signal data and perform waveform equalization processing without going through A / D conversion processing. Therefore,
It is possible to use a signal transmitted through a telephone line, a reproduction signal from an optical disc such as a CD (compact disc) and a signal obtained by converting an optical signal transmitted through a fiber into an electric signal by a photoelectric conversion element as input signal data. it can.

Further, the class dividing unit 4 generates a predetermined numerical value (class code) by performing binarization processing on the blocked signal data, but instead of this binarization processing, , ADRC (adaptive dynamic range encoding) processing may be used as the class code.

Further, although the class-based adaptive filter processing unit 5 performs processing using linear estimation coefficients, nonlinear quadratic estimation coefficients may be used instead of the linear estimation coefficients.

[0032]

As is apparent from the above description, the adaptive waveform equalizer according to the present invention divides the signal data transmitted or reproduced from the recording medium into blocks and class-divides them. A class dividing unit that converts the signal data of each block into a numerical value corresponding to the pattern of the signal data, and a filter coefficient from a table of a plurality of selectable filter coefficients stored in advance based on the converted numerical value. By including a table reading unit for reading, the signal data from the class dividing unit, and a sum of products processing unit for performing a sum of products operation using the filter coefficient read from the table reading unit, Since the number of signal states that can be handled can be increased, the performance of waveform equalization processing can be improved.

Further, the adaptive waveform equalization method according to the present invention is
A first step of dividing the signal data transmitted or reproduced from the recording medium into blocks and classifying, and converting the signal data of each class-divided block into a numerical value corresponding to the pattern of the signal data; and the above conversion. A second step of reading out the filter coefficient from a table of a plurality of selectable filter coefficients stored in advance based on the calculated numerical value, the signal data divided into the classes, and read from the table reading unit. By including the third step of performing the sum of products operation using the filter coefficient, the number of signal states that can be handled can be increased, so that the performance of waveform equalization processing can be improved.

[Brief description of drawings]

FIG. 1 is a diagram showing a schematic configuration of an adaptive waveform equalization device according to the present invention.

FIG. 2 is a diagram showing a schematic configuration of a class dividing unit 4 in FIG.

3 is a diagram showing a schematic configuration of a class-specific adaptive filter processing section 5 of FIG.

[Explanation of symbols]

3 ... Waveform equalization processing unit 4 ... Class division 5 ... Class-based adaptive filter processing unit 6 ...- Filter coefficient determination unit 9 ... Block processing unit 10 ... Threshold value determination processing unit 11 ... Binarization processing unit 12: Packing processing section 17: ROM table reading section 18: Linear estimation processing unit

─────────────────────────────────────────────────── ─── Continuation of the front page (56) Reference JP-A-4-259961 (JP, A) JP-A-55-109024 (JP, A) JP-A-60-139090 (JP, A) (58) Field (Int.Cl. ⁷ , DB name) H03H 21/00 H03H 17/00-17/08

Claims (6)

(57) [Claims] 1. The signal data transmitted or reproduced from a recording medium is divided into classes, and the class-divided signal data of each block is divided into classes .
Based on the class dividing unit that converts the numerical value corresponding to the pattern and the converted numerical value above, the pre-stored selection
A table of multiple selectable filter coefficients
Table reading unit for reading the number, the signal data from the class dividing unit, and the table
Using the filter coefficient read from the
An adaptive waveform equalization device having a product-sum processing unit for performing a product-sum operation 2. The filter coefficient read by the table reading unit is digital signal data before being transmitted or recorded on a recording medium and digital signal data after being transmitted or reproduced from the recording medium. respectively blocked before multivalued the digital signal data to be binarized, the adaptive waveform equalizer according to claim 1, wherein the obtained by learning using these blocked signal data 3. The class dividing unit includes a blocking processing unit for blocking multi-valued digital signal data into blocks, and a binarizing process for the blocked signal data from the blocking processing unit. A threshold value determination processing unit for determining a threshold value and a binarization for binarizing the blocked signal data from the blocking processing unit using the threshold value from the threshold value determination processing unit. 2. The adaptive waveform equalization apparatus according to claim 1, comprising a processing unit and a conversion processing unit for converting the binarized signal data for each block from the binarization processing unit into a corresponding numerical value. . 4. The signal data transmitted or reproduced from a recording medium is divided into classes and divided into classes , and the signal data of each of the divided classes is converted into the signal data of the signal data.
The first step of converting to the numerical value corresponding to the pattern , and the selection stored in advance based on the converted numerical value
The above filter from a table of possible multiple filter coefficients
The second step of reading the coefficient, the signal data divided into the classes , and the table
Sum of products using the filter coefficient read from the reading unit
An adaptive waveform equalization method , which comprises a third step of calculating 5. The filter coefficient read out in the table reading step includes digital signal data before being transmitted or recorded on a recording medium and digital signal data after being transmitted or reproduced from the recording medium. 5. The adaptive waveform equalization method according to claim 4 , wherein the multi-valued digital signal data before being digitized is divided into blocks, and learning is performed using these blocked signal data. 6. The class dividing step comprises a blocking processing step of dividing multivalued digital signal data into blocks.
Using a threshold value determining process step for determining a threshold value when binarizing the blocked signal data from the blocking process step, and the threshold value from the threshold value determining process step. And a binarization processing step of binarizing the blocked signal data from the block processing step,
5. The adaptive waveform equalization method according to claim 4 , further comprising a conversion processing step of converting the binarized signal data for each block from the binarization processing step into a corresponding numerical value.