JPH1051503A - Device and method for discriminating data - Google Patents

Abstract

PROBLEM TO BE SOLVED: To easily provide the error corresponding to the discriminated result, when discriminating data through a fixed delay tree search (FDTS). SOLUTION: When a stop length is defined as (k), at an FDTS device 1, the tree structure having the path of maximum path metric is selected and defined as the new object of FDTS between two tree structures, provided when the object tree structure of FDTS is divided into two at the first branching point and corresponding to the selected result, the discriminated result of data inputted just (k) samples preceding is outputted. Further, at the FDTS device 1, the errors respectively provided from 2<k> pieces of paths applied by the tree structure defined as the new object of FDTS are outputted and stored in a memory 21 . Corresponding to the discriminated result of data, selectors 31 -3k selectively output the half of errors stored in memories 21 -2k on the preceding stage.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to a data judging device and a data judging method, and more particularly to a data judging device and a data judging method for judging digital data transmitted from a transmission device or reproduced by a reproducing device, for example. .

[0002]

2. Description of the Related Art Conventionally, in a transmission apparatus for transmitting digital data and a reproduction apparatus for reproducing the digital data,
For example, to cope with interference and externally applied noise, digital data is equalized to desired characteristics,
The determination is performed on the equalized waveform (equalized data).

There are various methods for equalizing digital data. In the equalization, in particular, the equalization error included in the equalization waveform and the colorization of noise are determined by a highly reliable determination result. It needs to be kept to the extent that it can be obtained.

As a method of determining digital data, maximum likelihood decoding that traces a state transition by a code or TCM (Trelis Coding Modu) using a code with a large inter-code distance is used.
lation), and by using these, it is possible to reduce determination errors.

[0005] Further, as a determination method, an FDTS (Fixed Delay Tree Sea) for performing a maximum likelihood determination while performing equalization.
rch) (fixed delay tree search)) / DFE (Decision Feedba
ck Equalization). According to the DFE and the noise-whitening causalizing FFF including a determiner (hereinafter, appropriately referred to as an FDTS device) that makes a determination based on FDTS, it is possible to whiten noise and improve characteristics of the device with respect to the color of noise. . When an FDTS apparatus is used, the interference length of input data becomes very long. To cope with this, the FDTS apparatus is usually used in combination with a decision feedback equalizer.

[0006]

By the way, an equalization result corresponding to the digital data determined by the determiner, such as an equalization error or an equalized waveform, is input to another determiner (decoding device) or the like. In some cases, it is desired to obtain the determination result.

However, the FDTS / DFE device is
It is difficult to output the equalization result corresponding to the determined digital data because the FDTS device itself uses the equalization result corresponding to the digital data under determination for the determination, that is, a self-contained system. Met.

That is, in the FDTS apparatus, the equalization result corresponding to the data that has just been input is a so-called truncation length (tr
Since it is determined only after uncation length), it is difficult to output the equalization result.

The present invention has been made in view of such a situation, and when performing data determination by FDTS, it is possible to easily obtain an equalization error or an equalized waveform corresponding to the determination result. Is to be able to do.

[0010]

According to a first aspect of the present invention, there is provided a data determination apparatus, comprising: a maximum path metric of two tree structures obtained when a tree structure is divided into two at a first branch point; And choose the one with the path giving the new FDTS
And a tree structure selecting means for judging the data inputted k samples earlier in accordance with the selection result and outputting the judgment result, and a tree structure set as a target of the new FDTS. Output means for outputting predetermined path information obtained from each of the given 2 ^k paths;
An information storage / selection unit that stores 2 ^k pieces of path information output by the output unit and selects and outputs half of the path information according to the determination result.

According to a sixth aspect of the present invention, there is provided a data judging method in which a path having a maximum path metric is selected from two tree structures obtained when the tree structure is divided into two at the first branch point. Is selected as a target of a new FDTS, and in response to the selection result, data input k samples earlier is determined, the determination result is output, and a tree targeted as a new FDTS is output. 2 ^k given by the structure
Output predetermined path information obtained from each of the paths, store 2 ^k path information, and correspond to the determination result,
It is characterized in that half of the path information is selected and output.

In the data judging device according to the first aspect, the tree structure selecting means may select the largest one of the two tree structures obtained when the tree structure is divided into two at the first branch point. The one having a path that gives a metric is selected and set as a target of a new FDTS, and in accordance with the selection result, data input k samples earlier is determined.
The result of the determination is output. The output means outputs predetermined path information obtained from each of the 2 ^k paths provided by the tree structure targeted for the new FDTS, and the information storage and selection means outputs the 2 ^k paths output by the output means. The path information is stored, and half of the path information is selected and output according to the determination result.

According to a sixth aspect of the present invention, there is provided a data judging method comprising a path which gives a maximum path metric among two tree structures obtained when the tree structure is divided into two at the first branch point. Is selected and set as a new FDTS target, and in response to the selection result, the data input k samples earlier is determined, the determination result is output, and the new FDTS is determined. Output predetermined path information obtained from each of the 2 ^k paths given by the tree structure, store the 2 ^k path information, and select half of the path information according to the determination result. The output has been made.

[0014]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Embodiments of the present invention will be described below, but before that, in order to clarify the correspondence between each means of the invention described in the claims and the following embodiments. The features of the present invention are described as follows by adding a corresponding embodiment (however, an example) in parentheses after each means.

That is, the data determination device according to the first aspect of the present invention has a cutoff length of k for a tree structure that branches from one branch point and has two branches each assigned 0 or 1 as a minimum unit. FDTS (Fixed Delay Tr
ee Search), the 2k ^{+ 1} paths given by the tree structure are searched for those having the maximum likelihood with respect to the input data, and corresponding to the search result,
What is claimed is: 1. A data determination apparatus for determining data input k samples before, comprising: a branch metric calculation unit configured to calculate a branch metric of a k-th ^{2.sup.k + 1} branch in a tree structure using input data. (E.g., the branch metric calculator 11 shown in FIGS. 3 and 8) and the branch metric are used to provide a tree structure 2
^When the path metric calculation means (for example, the path metric update unit 12 shown in FIGS. 3 and 8) for calculating the path metric of ^{k + 1} paths, and when the tree structure is divided into two at the first branch point Out of the two tree structures obtained in step (a), the path having the path that gives the maximum path metric is selected and set as a target of a new FDTS, and according to the selection result, input is performed k samples earlier. Tree structure selecting means (for example, the selector 14 shown in FIGS. 3 and 8) for judging data and outputting the judgment result, and 2 ^k paths provided by the tree structure targeted for the new FDTS Output means (for example, the error memory 15 shown in FIG. 3 or the error memory 15 and the arithmetic unit 31 shown in FIG. 8) which outputs predetermined path information, and 2 ^k paths output by the output means. information Storing, in response to the determination result, information storage selection means for selecting a half of the path information (e.g., the memory 2 ₁ 2 _k 2 and 7
And characterized in that it comprises a selector 3 such ₁ to 3 _k) and.

[0016] data determination apparatus according to claim 4, the information storage selection means, the path information storage means for storing path information (e.g., a memory 2 ₁ 2 _k shown in FIGS. 2 and 7)
And path information selecting means (for example, selecting half of the path information stored in the path information in accordance with the determination result)
The combination of the selector 3 such ₁ to 3 _k) and shown in FIGS. 2 and 7, characterized in that it has only k stages.

According to a fifth aspect of the present invention, in the data judging device, the output means outputs errors corresponding to 2 ^k paths given by the tree structure selected by the tree structure selecting means as path information, and outputs k. An additional unit (for example, the computing unit 4 shown in FIG. 2) for adding the determination result to the error selected by the path information selecting unit at the stage is further provided.

Of course, this description does not mean that each means is limited to those described above.

Next, the FDTS to which the present invention is applied will be described.

[0020] As shown in FIG.
A path search is performed for a tree structure that has two branches, each of which is assigned a 0 or 1 as a minimum unit, which branches from a branch point (a portion indicated by a circle in the figure) of FIG. Assuming that the length (Truncation Length) is k, the depth (hierarchy) of the data input k samples earlier is determined in accordance with the search result of the path using the tree structure of k + 1. In FIG. 1, 1 or 0 is assigned to the upper or lower branch of the two branches extending from each branch point.

[0021] That is, for example, time t = n (where, n is an integer) in, if the data c _t is input, first, for the input data c _t, k-th from the head of the tree structure (tree It is calculated depth 2 ^{k + 1} pieces of branches each branch metric of the branch beginning of the structure to 0th), up to 2 ^k-number of branches each of the k-1 th depth, corresponding The path metric of the path up to each of the 2 ^{k + 1} branches at the k-th depth is calculated by adding the path metric to the path metric of the path.

Further, among the 2 ^{k + 1} paths, the path that gives the maximum path metric is detected, and the tree structure including the k-th branch is replaced with the first branch point (0-th 2nd path).
The branch having the path giving the maximum path metric is selected from the two tree structures obtained when the tree structure is divided into two branches. Then, the selected tree structure is shown in FIG.
Is the upper half or lower half indicated by U or D, the data input k samples earlier, that is, the data input at time t = nk, becomes the time t = n
It is determined that it is 1 or 0 assigned to the upper or lower branch of the two branches corresponding to -k, and the determination result is output.

Then, the same processing is repeated for the selected tree structure extended to the k-th.

Specifically, for example, at time t = n, assuming that the path indicated by the thick line in FIG. 1A gives the maximum path metric, the path including the path metric is included. In, the lower half tree structure indicated by D is selected. Further, in this case, the data determination result is 0
Is output. Then, at the next time t = n + 1, as shown in FIG. 1 (B), the same processing is performed on the selected lower half tree structure extended to the k-th.

According to the FDTS, as described above, each time data is input, the determination result of the data preceding the cutoff length k is sequentially output.

As for FDTS, for example, J. Mo
on, LRCarley, "Sequence Detection For High-Densi
The details are disclosed in “ty Storage Channels” and the like.

FIG. 2 shows the configuration of a first embodiment of the data judgment apparatus to which the present invention is applied. The FDTS apparatus 1, for example, magnetic or magneto-optical by a recording medium on which recording is performed (not shown) is reproduced from such data c _t
There have been made as input, wherein, by the FDTS as described above is carried out, k data c _t
The data c _tk before the sample is determined, and the determination result x
_tk is output. Determination result x _tk of the data is adapted to be supplied to the selector 3 ₁ to 3 _k and calculator 4.

Further, FDTS apparatus 1, when data c _t is input, it calculates the 2 ^{k + 1} pieces of branches each branch metric of k-th depth of the tree structure, demanded at that time, Data c _t corresponding to each of 2 ^{k + 1} paths
Error of the (path information), which corresponds to 2 ^k-number of path (2 ^k-number of errors set) is selected, is adapted to be supplied to the memory 2 _1.

Memory 2 _i (i is an integer of 1 or more and k or less)
Are provided for the number of stages equal to the cutoff length k of the FDTS performed by the FDTS device 1, and store the output of the selector 3 _{i-1 at} the preceding stage. Note that only the memory 2 _1, as described above, is adapted to store the output of the FDTS device 1. The selector 3 _i is an FDTS device 1
Corresponding to the determination result x _tk from the upper half U or the lower half of the stored value of the memory 2 _i ,
The selected value is output to the memory 2 _{i + 1} at the subsequent stage. Note that only the selector 3 _k outputs the selected value as
The data is output as an error of the data _ctk input k samples before. The error output from the selector 3 _k is also supplied to the arithmetic unit 4.

The arithmetic unit 4 adds the judgment result x _tk from the FDTS device 1 and the error from the selector 3 _k , and equalizes the addition result, ie, the data c _tk input k samples before. A waveform (the maximum likelihood waveform) is output.

[0031] Note that the FDTS apparatus 1 and the memory 2 ₁ to 2 _k, have been made so that the clock is supplied from the clock generation circuit (not shown), FDTS apparatus 1 and the memory 2 ₁ 2 _k, the clock It is made to operate in synchronization with.

FIG. 3 shows an example of the configuration of the FDTS apparatus 1 shown in FIG. Input data c _t is adapted to be supplied to the branch metric calculation unit 11, the branch metric calculation unit 11 for the data c _t, k-th 2 ^{k +1} amino branch branch metrics in the tree structure Each is calculated, and the path metric updating unit 12
To be supplied.

Here, the branch metric is, for example,
As disclosed on page 77 of the above-mentioned document, it is calculated according to the following equation. That is, the k-th 2 in the tree structure
^When the branch metric of ^{k + 1} branches is λ _j (where j is an integer of 0 or more and less than 2 ^{k + 1} ), and when the reference waveform sequence is y _j , the branch metric λ _j is calculated by the following equation. Is done.

The _{λ j = - (c t -y} j) 2 ··· (1) Here, the reference waveform sequence y _j, recording and reproducing system impulse response of the data determination unit of FIG. 2 is used (data determination If the device is used in other systems, such as transmission systems,
K-th order FIR with the impulse response of the system) as a coefficient
(Finite Impulse Response) The FIR obtained when the value (0 or 1) assigned to each branch constituting the path up to the k-th 2 ^{k + 1} branches in the tree structure is input to the filter. Means the output of the filter.

Further, the branch metric calculator 11
Is obtained when calculating the 2 k ^{+ 1} pieces of branches each branch metric corresponding to 2 k ^{+ 1} single pass, and stores and supplies to the error memory 15 an error n _j data c _t It has been done as well. That is, the error n _j is given by the formula n _j = c _t -y _j, which is obtained when calculating the formula (1). Branch metric calculator 11
Supplies the error n _j obtained in this way to the error memory 15 for storage.

The path metric updating unit 12 reads the path metrics of 2 ^k paths up to each of the (k−1) -th branches in the tree structure from the path metric memory 13, and stores the path metrics into the 2 ^k path metrics. , By adding the corresponding branch metrics from the branch metric calculation unit 11 to the k-th branch at each depth.
The path metric of ^{k + 1} paths is calculated (the path metric is updated). The path metrics of the 2 ^{k + 1} paths are supplied from the path metric updating unit 12 to the path memory 13.

The path memory 13 stores 2 ^{k + 1} path metrics supplied from the path metric updating unit 12.

The selector 14 detects (searches) the maximum value from the 2 ^{k + 1} path metrics stored in the path memory 13 and, as described with reference to FIG. Is divided into two tree structures at the first branch point, and a tree structure including a path that gives the maximum path metric is selected.

Further, the selector 14 responds to the selection result by inputting the data c _tk input k samples earlier.
The outputs the determination value x _tk, of the 2 ^{k + 1} single path metric stored in the path metric memory 13, leaving only those paths belonging to the selected tree structure,
The other 2 ^k path metrics are deleted. Here, the remaining 2 ^k path metrics are
Used to calculate the next path metric.

Further, the selector 14 correspond to the respective 2 ^{k + 1} single path stored in the error memory 15, of error n _j data c _t, the 2 ^k-number of paths belonging to the selected tree structure The corresponding thing is also selected.

The error memory 15 stores 2 ^{k + 1} errors n _j supplied from the branch metric calculation unit 11 and, out of the stored errors n _j , 2 ^k errors n _j selected by the selector 14 and it is adapted to output to the memory 2 ₁ (FIG. 2).

Next, the operation of the data judging device of FIG. 2 will be described with reference to FIG. 4 and FIG. In the data determination device, first, in step S1 of FIG. 4, a variable t representing time is initialized to, for example, 0,
Proceeding to 2, the initial values of the path metrics up to the (k-1) -th 2 ^k branches in the tree structure are set. As the initial value, for example, 0 or the like is used, and the setting of the initial value is performed by, for example, the path metric updating unit 12 or the like.

The initial value of the path metric set in step S2 is stored in the path metric memory 13 (FIG. 3).
And stored. When the data c _t is input, in step S3, the branch metric calculation unit 11 for the data c _t, k-th 2 ^{k + 1} pieces of branches each branch metric in the tree structure lambda _j
Is calculated and supplied to the path metric updating unit 12. Further, in step S3, the branch metric calculation unit 11 calculates an error (equalization error, noise component included in data, etc.) corresponding to each of the 2 ^{k + 1} paths obtained when calculating the branch metric λ _j. ) _Nj is supplied to the error memory 15 for storage.

The path metric updating unit 12 outputs
The branch metric is received from the trick calculator 11.
Then, in step S4, the k-1 th in the tree structure
2 to each branch at eye depth^kPath metrics for paths
Block is read from the path metric memory 13 and
From the corresponding branch
By adding tricks, each branch at the k-th depth
Up to 2^{k + 1}Compute the path metric for this path. Sa
In addition, the path metric updating unit 12^{k + 1}Pieces
Path metric is supplied to the path memory 13 and stored.
By doing so, the stored value of the path memory 13 is updated.

When the update of the path memory 13 is performed, the selector 14 detects the maximum path metric stored in the path memory 13 in step S5, and the tree structure including the path giving the path metric is shown in FIG. It is determined whether the tree structure is a tree structure (tree group) on the upper U side shown in FIG. 1A or a tree structure on the lower D side.

If it is determined in step S5 that the tree structure including the maximum path metric is on the upper U side, the process proceeds to step S6, where the selector 14 determines the path metric of the path given by the upper U side tree structure. Is selected, and the path metric of the path given by the remaining tree structure on the lower D side is deleted from the path memory 13. Further, in step S7, the selector 14 outputs 1 as the judgment value (judgment result) _xtk , and in step S1
Go to 0.

On the other hand, if it is determined in step S5 that the tree structure including the maximum path metric is on the lower D side, the process proceeds to step S8, where the selector 14
Only the path metric of the path provided by the tree structure on the side is selected, and the path metric of the path provided by the tree structure on the remaining upper U side is deleted from the path memory 13.
Further, in step S9, the selector 14 outputs 0 as the determination value _xtk , and proceeds to step S10.

In step S10, the selector 14
Corresponds to a respective 2 ^{k + 1} single path stored in the error memory 15, of error n _j data c _t, those corresponding to the 2 ^k-number of paths belonging to the tree structure that selects the path metric Then, the selected 2 ^k errors n _j are output from the error memory 15.

The error n _j of the 2 ^k pieces is stored is supplied to the memory 2 ₁ (FIG. 2). Here, the memory 2 _1, of the error n _j corresponding to the path to the FIG. 1 k-th tree structure shown in (A), 2 is arranged longitudinally ^{k + 1} single branch respectively 2 ^k selected by the selector 14
Are stored in such a state that they are arranged in the vertical direction.

After the processing in step S10, step S1
Proceeding to 1, the variable t is incremented by 1 and the process returns to step S3.

Meanwhile, the selector 14, and outputs a determination value x _tk at step S7 or S9, the selector 3 ₁ to 3 _k (FIG. 2), in step S16 of FIG. 5,
It is determined whether the determination value _xtk is 0 or 1. In step S16, the determination value x _tk
Is determined to be 1, the process proceeds to step S17,
Selector 3 _1, stored in the memory 2 _1, 2 of the ^k path error n _j corresponding to each select an error n _j of stored halved 2 ^k-1 pieces of the upper U-side and, stores and supplies to the subsequent memory 2 _2, and returns. Further, in step S16, if the determination value x _tk is determined to be 0, the process proceeds to step S18, the selector 3 _1, stored in the memory 2 _1, error n corresponding to each 2 ^k-number of path of _j, select 2 ^k-1 pieces of error n _j stored in one half of the lower side D, also, stores and supplies to the subsequent memory 2 _2, and returns.

Similar processing is performed in the selectors 3 _{2 to} 3 _k−1 , whereby 2 ^{k−i + 1} errors are stored in the memory 2 _i . Therefore, two errors are stored in the memory 2 _k , and one of the two errors is stored in the selector 3 _k.
And output as an error k samples earlier.

As described above, in the selectors 3 _{1 to} 3 _k , the error obtained on the upper or lower side of the tree structure that includes the path giving the maximum path metric corresponding to the input data is obtained. Are sequentially selected, and finally, the data c input k samples earlier
The error of _tk is output.

This error is supplied to the arithmetic unit 4,
By being added to the judgment value _xtk , the data is output as an equalized waveform of the data _ctk .

As described above, the memory 2 _i has
Since 2 ^{k−i + 1} errors are stored, memories 2 _{1 to} 2 _k
The total storage capacity of the first term and the common ratio are both 2
Where the sum of geometric progressions whose number of terms is equal to the truncation length k, ie
It suffices if there is an amount that can store 2 ^{k + 1} -2 errors.

Next, FIG. 6 shows that the cutoff length k is, for example, 2
3 shows a configuration example of the data determination device of FIG. In this case, since the cutoff length k is 2, the FDT
Every time the tree structure is selected by the selector 14 of the S device 1, 4 (= 2 ² ) errors are output from the FDTS device ₁ and stored in the memory 21.

Meanwhile, the selector 3 _1, corresponding to the determined value x _t-2, of the four error stored in the memory 2 _1,
Is the upper half or the lower half 2 is selected, is supplied to and stored in the memory 2 _2. At the same time, the selector 3 _2, in response to the determination value x _t-2, stored in the memory 2 ₂ 2
One of the upper half or lower half of the errors is selected and output as an error with respect to data two samples before. Further, the arithmetic unit 4, the determination value x _t-2, is added to the output of the selector 3 _2, is output as an equalized waveform for two samples previous data.

Next, FIG. 7 shows the configuration of a second embodiment of the data judgment apparatus to which the present invention is applied. In the figure,
Parts corresponding to those in FIG. 2 are denoted by the same reference numerals. That is, the data determination device uses the FDT
An FDTS device 2 is provided instead of the S device 1, and
The configuration is the same as that of the data determination device in FIG. 2 except that the arithmetic unit 4 is omitted.

Here, the data judgment device of FIG. 2 outputs an error and an equalized waveform together with the judgment value _xtk , but the data judgment device of FIG.
Along with _tk , only the equalized waveform is output.

FIG. 8 shows an example of the configuration of the FDTS device 21 of FIG. Note that, in the figure, parts corresponding to those in FIG. 3 are denoted by the same reference numerals. That is, F
The DTS device 21 has the same configuration as the FDTS device 1 except that an arithmetic unit 31 is newly provided. The arithmetic unit 31 calculates a determination value _xtk output from the selector 14,
An equalized waveform is obtained and output by adding the error n _j output from the error memory 15.

Next, the operation of the data judging device of FIG. 7 will be described with reference to FIG. In this data determination device, in steps S21 to S29, the same processing as in steps S1 to S9 of FIG. 4 is performed. Then, after the processing of step S27 or S29,
Proceeding to step S30, first, as in step S10 of FIG.
corresponding to ^{k + 1} single pass, error n _j data c _t
Of the tree structures corresponding to 2 ^k paths belonging to the selected one of the upper and lower tree structures are selected by the selector 14.
And the selected 2 ^k errors n _j are
It is output to the arithmetic unit 31 (FIG. 8).

[0062] The arithmetic unit 31, in addition to the output of the error memory 15, also the determination value _xk of the selector 14 outputs are supplied, the calculator 31, by which both are added, the data c _t, selection 2 ^k-number of paths corresponding to the equalized waveform (path information) is required to belong to have been a tree structure, it is supplied to the memory 2 _1.

The selectors 3 _{1 to} 3 _k in FIG. 7 perform the same processing as the processing described with reference to FIG. 5 on the equalized waveform, not on the error. An equalized waveform (maximum likelihood waveform) of the data _ctk is output.

According to the data determination device of FIG. 7 as described above, the arithmetic unit 4 (FIG. 2) does not need to be provided as an external block of the FDTS device 21.

Next, FIG. 10 shows an example of the configuration of the data judging device of FIG. 7 when the censoring length k is 2, for example. In the figure, the same reference numerals are given to portions corresponding to the case in FIG.

In this case, since the cutoff length k is 2,
Each time the tree structure is selected in the selector 14 of the FDTS device 21, the FDTS device 21 outputs 4 (=
2 ²⁾ pieces of equalized waveform is output and stored in the memory 2 _1.

[0067] Hereinafter, by processing similar to that in FIG. 6 is performed, the selector 3 _2, equalized waveform for the two samples previous data is output.

As described above, while outputting the error or equalized waveform as the predetermined path information obtained from each of the k-th 2 ^k paths in the tree structure targeted for the new FDTS, The information is stored, and in accordance with the data determination result, the process of selecting and outputting half of the stored path information is repeated.
When data is determined by the TS, an error or an equalized waveform corresponding to the determination result can be easily output.

As described above, these errors and equalized waveforms can be given as inputs to an external device. For example, a decision feedback equalizer such as an adaptive equalized type It can be used as information for updating the tap coefficient of the object.

That is, the FDTS device is usually used as a decoding device combined with a decision feedback equalizer. In this case, the decoding device is configured, for example, as shown in FIG. That is, the data is an FFF (Feed Forward Filtration).
ter) 41, filtered, and supplied to a computing unit 42. The arithmetic unit 42 is supplied with the output of the FBF (Feed Back Filter) 44 in addition to the output of the FFF 41, where the difference between the output of the FFF 41 and the output of the FBF 44 is calculated. The difference value is supplied to the data determination device 43. Data judgment device 4
Reference numeral 3 denotes an FDTS device that performs determination based on FDTS, performs processing (FDTS) on the output of the arithmetic unit 42, and outputs the data determination result to the FBF 44. In the FBF 44, the determination result from the data determination device 43 is filtered and supplied to the arithmetic unit 42.

In the above decoding device, the FFF
41 and the FBF 44 function as equalizers. These are used as adaptive equalizers, and the data determination device 43 is configured as shown in FIG. 2 or FIG.
By supplying the error or equalized waveform obtained by the data determination device 43 to the FF 41 and the FBF 44, the FFF 4
In 1 and the FBF 44, an optimal tap coefficient can be set.

The data determination device described above can be applied to, for example, a magnetic disk device, a magneto-optical disk device, a magnetic tape device, and a transmission device for transmitting data reproduced from those devices. is there.

Further, in this embodiment, the FDTS apparatus 1 (or 21) selects a tree structure including the maximum path metric, and selects an error (or equalization) corresponding to 2 ^k paths given by the tree structure. Waveform)), but only the error corresponding to the path giving the maximum path metric is output, and the error is delayed by k samples and input before k samples. It is also possible to output as an error (approximate value) of the obtained data. In this case, the memories 2 _{1 to} 2 _k
The total storage capacity of, requires if by the amount for storing k number of the error, also, the selector 3 ₁ to 3 _k need not be provided.

Further, in this embodiment, although not particularly mentioned, a sequence of 0s or 1s as input data,
The so-called d limit (run limit) (Minimam Run Length Lim
ited) is specified, and if there is a sequence of data that cannot appear, from the tree structure targeted for FDTS,
It is possible to exclude such a sequence of data and corresponding paths. In this case, the memories 2 _{1 to} 2 _k
Can be reduced.

[0075]

According to the data judging device according to the first aspect and the data judging method according to the sixth aspect, the tree structure is
Of the two tree structures obtained when the tree is divided into two at the first branch point, the one having the path that gives the maximum path metric is selected and set as the target of a new FDTS. Correspondingly, the data input k samples earlier is determined, and the determination result is output.
Further, predetermined path information obtained from each of the 2 ^k paths provided by the tree structure targeted for the new FDTS is output, and the 2 ^k path information is stored. Then, according to the determination result of the output data,
Half of the path information is selected and output. Therefore, when data is determined by FDTS,
It is possible to easily output path information such as an error and an equalized waveform corresponding to the determination result.

[Brief description of the drawings]

FIG. 1 is a diagram for explaining FDTS.

FIG. 2 is a block diagram illustrating a configuration of a first embodiment of a data determination device to which the present invention has been applied;

FIG. 3 is a block diagram illustrating a configuration example of the FDTS device 1 of FIG. 2;

FIG. 4 is a flowchart illustrating a process of the data determination device of FIG. 2;

FIG. 5 is a flowchart illustrating a process of the data determination device of FIG. 2;

FIG. 6 is a block diagram showing a configuration example of the data determination device of FIG. 2 when the cutoff length k is 2.

FIG. 7 is a block diagram illustrating a configuration of a second embodiment of the data determination device to which the present invention has been applied.

8 is a block diagram illustrating a configuration example of the FDTS device 21 of FIG.

FIG. 9 is a flowchart illustrating a process of the data determination device of FIG. 7;

FIG. 10 is a block diagram illustrating a configuration example of the data determination device in FIG. 7 when the cutoff length k is 2.

FIG. 11 is a block diagram illustrating a configuration example of a decoding device in which an FDTS device and a decision feedback equalizer are combined.

[Explanation of symbols]

1 FDTS device, 2 _{1 to} 2 _k memory, 3 _{1 to} 3 _k selector, 4 arithmetic unit, 11 branch metric calculation unit, 12 path metric update unit, 13 path metric memory, 14 selector, 15 error memory, 21 FDTS device, 31 arithmetic unit, 41
FFF, 42 operation unit, 43 data judgment device,
44 FBF

Claims (6)

[Claims] An FDTS (Fixed) is set for a tree structure having two branches, each of which is assigned a 0 or 1 as a minimum unit, which branches from a branch point of 1 and has a cutoff length of k.
By performing Delay Tree Search), a search is performed for a path having the maximum likelihood with respect to input data, out of 2 ^{k + 1} paths provided by the tree structure, and k is determined according to the search result. What is claimed is: 1. A data determination apparatus for determining data input before a sample, comprising: calculating a branch metric of a k-th 2 ^{k + 1} branches in the tree structure using the input data. Means, using the branch metric, a path metric calculation means for calculating a path metric of 2 ^{k + 1} paths given by the tree structure, and dividing the tree structure into two at the first branch point Of the two tree structures that are sometimes obtained, the one having the path that gives the maximum path metric is selected and set as the target of a new FDTS, and according to the selection result, Determining the input data samples just before, the tree structure selection means for outputting a result of the determination, obtained from 2 ^k-number of paths respectively provided by the tree structure that is the subject of a new FDTS, predetermined An output unit that outputs path information; an information storage selection that stores 2 ^k pieces of the path information output by the output unit and that selects and outputs half of the path information according to the determination result And a means for determining data. 2. The method according to claim 1, wherein the output unit outputs an error corresponding to each of 2 ^k paths provided by the tree structure selected by the tree structure selection unit as the path information. 2. The data determination device according to 1. 3. The output means adds the determination result to an error corresponding to each of 2 ^k paths provided by the tree structure selected by the tree structure selection means,
The data determination device according to claim 1, wherein the added value is output as the path information. 4. The information storage and selection unit includes: a path information storage unit that stores the path information; and a path that selects half of the path information stored in the path information according to the determination result. 2. The data judging device according to claim 1, wherein the data judging device has only k stages in combination with the information selecting means. 5. The output means outputs, as the path information, errors corresponding to 2 ^k paths provided by the tree structure selected by the tree structure selection means as the path information. 5. The data determination device according to claim 4, further comprising an addition unit that adds the determination result to the error selected by the information selection unit. 6. An FDTS (Fixed) method for a tree structure that branches from a branch point of 1 and has two branches each assigned 0 or 1 as a minimum unit, where k is a truncation length.
Delay Tree Search), a search is performed for a path having the maximum likelihood with respect to the input data, out of 2 ^{k + 1} paths provided by the tree structure, and k is determined according to the search result. A data determination method for determining data input just before a sample, comprising calculating a branch metric of the k-th 2 ^{k + 1} branches in the tree structure using the input data, Using a metric, a path metric of 2 ^{k + 1} paths given by the tree structure is calculated. Of the two tree structures obtained when the tree structure is divided into two at the first branch point, Of the path that gives the maximum path metric is selected as a target of a new FDTS, and data input k samples before is determined according to the selection result. Output obtained from each 2 ^k-number of paths provided by the tree structure that is the subject of new FDTS, outputs a predetermined path information, stores 2 ^k pieces the path information, the determination result Correspondingly, a half of the path information is selected and outputted.