JP6215271B2 - Signal processing apparatus, magnetic information reproducing apparatus, and signal processing method - Google Patents

Description

  The present invention relates to a signal processing device, a magnetic information reproducing device, and a signal processing method.

  Patent Document 1 discloses a technique for adjusting the phase of a signal input to the input stage of the phase adjustment circuit by feeding back the output of the decoder to the input stage of the phase adjustment circuit. Further, in the technique described in Patent Document 1, an optimum phase feedback circuit according to a decoded pattern is enabled by mounting an independent phase adjustment circuit for each possible decoding pattern.

  Non-Patent Document 1 discloses a technique for minimizing the feedback delay by optimizing the gain parameter of the phase adjustment circuit even in a system in which there is a feedback delay to the input stage of the phase adjustment circuit. .

  In Patent Document 2, since a plurality of channels are simultaneously affected by speed fluctuations in the tape system, not only a single channel signal but also information on speed fluctuations obtained from reproduction signals of other channels is used. A method for improving tolerance to jitter is disclosed.

US Pat. No. 7,602,863 B2 US Patent 7885030B2

J.Xie and B.V.K.V.Kumar, "Timing Recovery Loop Delay Compensation by Optimal Loop Gains",; in Proc.ICC, 2006, pp.3229-3234

  However, with the technique described in Patent Document 1, since the final output of the decoder is fed back, it is difficult to immediately adjust the phase of the signal input to the input stage of the phase adjustment circuit. Although it is possible to suppress a drop in immediacy by installing a plurality of phase adjustment circuits and a plurality of buffer memories, the circuit scale becomes enormous.

  Further, in the technique described in Non-Patent Document 1, since the reliability of the signal fed back to the input stage of the phase adjustment circuit is low, the phase adjustment circuit does not function properly.

  The present invention has been proposed in view of such circumstances, and a specific decoding target is compared with a case where the decoding target signal itself is used for adjustment of extraction timing and a case where the final decoding result is used for adjustment of extraction timing. It is an object of the present invention to provide a signal processing device, a magnetic information reproducing device, and a signal processing method that can achieve both high-accuracy signal extraction and suppression of delay caused by adjustment of extraction timing.

  To achieve the above object, the signal processing apparatus according to the first aspect of the present invention extracts an extraction target signal from an input digital signal at an extraction timing determined as a timing for extracting the decoding target signal. A decoding unit that decodes the decoding target signal by estimating the decoding result candidate of the decoding target signal extracted by the extraction unit by the maximum likelihood method and detecting the maximum likelihood decoding result, and the decoding unit And an adjustment unit that adjusts the extraction timing using the likelihood of the candidate decoding result. Thereby, the signal processing apparatus can extract a specific decoding target signal with high accuracy and the extraction timing compared to the case where the decoding target signal itself is used to adjust the extraction timing and the case where the final decoding result is used to adjust the extraction timing. It is possible to achieve both suppression of delay caused by adjustment.

  In the signal processing device according to the second aspect of the present invention, the decoding unit estimates a candidate by the maximum likelihood method and detects the maximum likelihood decoding result, and the maximum likelihood decoding result detected by the detection unit And a correction unit for correcting the error. Thereby, the signal processing apparatus can suppress the delay caused by the adjustment of the extraction timing, compared to the case where the extraction timing is adjusted based on the output of the correction unit.

  In the signal processing device according to the third aspect of the present invention, the adjustment of the extraction timing by the adjustment unit is determined in advance as an upper limit time in which the deviation of the extraction timing is allowed after the extraction unit starts extracting the decoding target signal. It starts on the condition that the elapsed time has passed. Thereby, the signal processing apparatus is compared with the case where the adjustment of the extraction timing is performed before the predetermined time has elapsed as the upper limit time in which the deviation of the extraction timing is allowed after the extraction of the decoding target signal is started. The reliability of extraction timing can be increased.

  In the signal processing device according to the fourth aspect of the present invention, the adjustment unit extracts the decoding target signal by the extraction unit after a predetermined time has elapsed since the extraction unit started extracting the decoding target signal. Each time, the likelihood based on the extracted decoding target signal is calculated, and each time the likelihood is calculated, the extraction timing is adjusted using the calculated likelihood. As a result, the signal processing apparatus can perform the extraction timing adjustment using the likelihood first calculated after a predetermined time has elapsed since the extraction of the decoding target signal is started only once. In comparison, even if the decoding target signal fluctuates, high reliability with respect to the extraction timing can be ensured.

  In the signal processing device according to the fifth aspect of the present invention, each time the extraction target signal is extracted by the extraction unit, the adjustment unit calculates a likelihood based on the extracted decoding target signal, and calculates the likelihood. Each time, the extraction timing is adjusted using the calculated likelihood. Thereby, the signal processing apparatus can suppress that the adjustment precision of extraction timing falls with the fluctuation | variation of a decoding target signal compared with the case where adjustment of extraction timing using likelihood is performed only once. .

  In the signal processing device according to the sixth aspect of the present invention, the adjustment unit extracts each time the signal to be decoded is extracted by the extraction unit, regardless of whether the maximum likelihood decoding result is detected by the decoding unit. The likelihood based on the decoded decoding target signal is calculated, and each time the likelihood is calculated, the extraction timing is adjusted using the calculated likelihood. As a result, the signal processing apparatus can suppress a delay that occurs due to the adjustment of the extraction timing, compared to the case where the extraction timing is adjusted after waiting for the maximum likelihood decoding result to be detected.

  In the signal processing device according to the seventh aspect of the present invention, the adjustment unit decreases the adjustment amount of the extraction timing as the likelihood is lower. Thereby, the signal processing apparatus can improve the adjustment accuracy of the extraction timing as compared with the case where the adjustment amount of the extraction timing is determined regardless of the likelihood.

  In the signal processing device according to the eighth aspect of the present invention, the likelihood is a likelihood calculated using soft information generated based on the decoding target signal. Thereby, the signal processing apparatus can obtain a highly accurate likelihood compared with the case where the likelihood is defined without using soft information.

  In the signal processing device according to the ninth aspect of the present invention, the likelihood is defined by the probability that the decoding result is correctly detected by the decoding unit, and the probability uses the maximum likelihood path metric and the competitive path metric that are soft information. Is the calculated probability. Thereby, the signal processing apparatus can obtain a highly accurate likelihood as compared with the case where the likelihood is defined only by the maximum likelihood path metric or the competitive path metric.

In the signal processing device according to the tenth aspect of the present invention, the adjustment unit sets k as a natural number, ε as a phase error between the decoding target signal and an ideal decoding target signal, and y as a signal level of the decoding target signal. , D is the ideal signal level of the decoding target signal, τ is the adjustment amount of the extraction timing, g is the probability, θ is the differential term of the extraction timing error, and α and β are the adjustment terms below. The extraction timing is adjusted using the adjustment amount τ _{k + 1} obtained by the mathematical formulas (1), (2), and (3). Thereby, the signal processing device can adjust the extraction timing with higher accuracy than the case where the extraction timing is adjusted only by the phase error calculated using Expression (1).

ε _k ≡y _k d _k−1 −y _k−1 d _k (1)
θ _k = θ _k−1 + g _k βε _k (2)
τ _{k + 1} = τ _k + g _k αε _k + θ _k (3)

  In the signal processing device according to the eleventh aspect of the present invention, the maximum likelihood method is a maximum likelihood method based on the Viterbi algorithm. Thereby, the signal processing apparatus can obtain a high cost-effectiveness compared with the case where the maximum likelihood method using an algorithm other than the Viterbi algorithm is adopted.

  The signal processing device according to the twelfth aspect of the present invention further includes an amplitude component adjusting unit that adjusts the amplitude component of the decoding target signal, and the decoding unit is configured to output the decoding target signal whose amplitude component has been adjusted by the amplitude component adjusting unit. The candidate of decoding result is estimated by the maximum likelihood method, and the maximum likelihood decoding result is detected. Thereby, the signal processing apparatus can improve the detection accuracy of the maximum likelihood decoding result by the decoding unit as compared with the case where the decoding target signal extracted by the extracting unit is directly input to the decoding unit.

  A magnetic information reproducing apparatus according to a thirteenth aspect of the present invention is generated by a read head that reads magnetic information from a magnetic recording medium, a generation unit that generates a digital signal from magnetic information read by the read head, and a generation unit. And a signal processing device according to any one of the first to twelfth aspects. Thus, the magnetic information reproducing apparatus can extract and extract a specific decoding target signal with higher accuracy than when the decoding target signal itself is used for adjustment of the extraction timing and when the final decoding result is used for adjustment of the extraction timing. It is possible to achieve both the suppression of the delay caused by the adjustment.

  In the magnetic information reproducing apparatus according to the fourteenth aspect of the present invention, the magnetic recording medium is a magnetic tape. Thereby, in the magnetic information reproducing apparatus, a change in the transport speed of the magnetic tape occurred locally as compared with the case where the decoding target signal itself is used for adjusting the extraction timing and the case where the final decoding result is used for adjusting the extraction timing. Even in this case, it is possible to achieve both high-precision extraction of a specific decoding target signal and suppression of delay caused by adjustment of extraction timing.

  In the magnetic information reproducing device according to the fifteenth aspect of the present invention, the adjustment of the extraction timing by the adjusting unit included in the magnetic information reproducing device starts with the extraction of the decoding target signal by the extracting unit included in the magnetic information reproducing device. To be started on the condition that a predetermined time has elapsed as an allowable time for jitter generated with the conveyance of the magnetic tape. The magnetic information reproducing apparatus can increase the reliability of the extraction timing as compared with the case where the extraction timing is adjusted on the condition that the extraction of the decoding target signal is started.

  The signal processing method according to the sixteenth aspect of the present invention extracts a decoding target signal from an input digital signal at an extraction timing determined as a timing for extracting the decoding target signal, and decodes the extracted decoding target signal. And decoding the decoding target signal by detecting the maximum likelihood decoding result and adjusting the extraction timing using the estimated likelihood of the decoding result candidate. As a result, the signal processing method has a higher accuracy of extraction and extraction timing of a specific decoding target signal than when the decoding target signal itself is used for adjustment of extraction timing and when the final decoding result is used for adjustment of extraction timing. It is possible to achieve both suppression of delay caused by adjustment.

  According to the present invention, compared with the case where the decoding target signal itself is used for adjusting the extraction timing and the case where the final decoding result is used for adjusting the extraction timing, the specific decoding target signal can be extracted with high accuracy and the extraction timing can be adjusted. Therefore, it is possible to achieve both the suppression of the delay caused by this.

It is a block diagram which shows an example of the principal part structure of the magnetic information reproducing | regenerating apparatus which concerns on embodiment. It is a flowchart which shows an example of the flow of the sampling process which concerns on embodiment. It is a flowchart which shows an example of the flow of the Viterbi detection process which concerns on embodiment. It is a flowchart which shows an example of the flow of the sampling time adjustment process which concerns on embodiment. It is a graph which shows an example of the correspondence of probability _gk and an error occurrence location. An example of the total number of error bits (broken line) with respect to the processing data length when the probability g _{k represented} by the equation (4) is used, and the total number of error bits (solid line) with respect to the processing data length when the probability g _k is a fixed value It is a graph to show. Processing data length when SNR (signal-noise ratio) (broken line) and probability g _k are fixed values with respect to processing data length when probability g _k shown by equation (4) is used It is a graph which shows an example with respect to SNR (solid line). An example of a phase error (broken line) with respect to the processing data length when the probability g _{k represented} by the mathematical formula (4) is used and a phase error (solid line) with respect to the processing data length when the probability g _k is a fixed value are shown. It is a graph. FIG. 7 is a graph showing an example of an SNR (broken line) with respect to the processing data length when the probability g _{k represented} by the equation (4) is used, and an SNR (solid line) with respect to the processing data length when the probability g _k is a fixed value. is there. An example of a phase error (broken line) with respect to the processing data length when the probability g _{k represented} by the mathematical formula (4) is used and a phase error (solid line) with respect to the processing data length when the probability g _k is a fixed value are shown. It is a graph. It is a block diagram which shows the modification of the principal part structure of the magnetic information reproducing | regenerating apparatus which concerns on embodiment. It is a block diagram which shows the modification of the principal part structure of the magnetic information reproducing | regenerating apparatus which concerns on embodiment. It is a block diagram which shows an example of a principal part structure of the magnetic information reproducing | regenerating apparatus based on 1st prior art. It is a block diagram which shows an example of a principal part structure of the magnetic information reproducing | regenerating apparatus based on 2nd prior art.

  Hereinafter, an example of an embodiment of a magnetic information reproducing apparatus according to the present invention will be described with reference to the accompanying drawings.

  As an example, as shown in FIG. 1, the magnetic information reproducing apparatus 10 includes a magnetic head 12, an A / D (analog / digital) converter 14, an equalizer 16, a sampling circuit 18, a decoding unit 20, a computer 22, and an adjustment. A value calculation circuit 24 is included.

  The magnetic head 12 is an example of a read head according to the present invention, and the A / D converter 14 and the equalizer 16 are examples of a generating unit according to the present invention. The sampling circuit 18 is an example of an extraction unit and an adjustment unit according to the present invention, and the adjustment value calculation circuit 24 is an example of an adjustment unit according to the present invention.

  The magnetic head 12 reads magnetic information from a magnetic tape 26 (an example of a magnetic recording medium according to the present invention) being conveyed in the direction of arrow A, and converts an analog signal corresponding to the read magnetic information to an A / D converter. 14 for output.

  The A / D converter 14 converts the analog signal input from the magnetic head 12 into a digital signal at a specific period, and outputs the digital signal obtained by the conversion to the equalizer 16. Here, in the present embodiment, the specific period refers to a period of 10 nanoseconds defined by the system clock, for example.

  The equalizer 16 performs waveform filtering on the time series data by performing digital filtering on the time series data that is the time series digital signal input by the A / D converter 14. That is, the equalizer 16 shapes the time-series data so that the decoding target signal, which is a digital signal to be decoded by the decoding unit 20, is sampled by the sampling circuit 18 and output to the decoding unit 20. Then, the equalizer 16 outputs a digital signal included in the time series data obtained by waveform equalization to the sampling circuit 18 at a specific period.

  The sampling circuit 18 samples the decoding target signal from the digital signal input from the equalizer 16 at a specific sampling period, and outputs the decoding target signal to the decoding unit 20 at a specific output period. Here, the specific sampling period refers to a period determined as a period for specifying an interval in which the decoding target signal exists in the digital signal input from the equalizer 16, and the specific output period is specified. Refers to the period synchronized with the sampling period.

  Note that the sampling timing of each decoding target signal by the sampling circuit 18, that is, the sampling timing, is adjusted by the sampling circuit 18 and the adjustment value calculation circuit 24. The adjustment of the sampling time is realized by adjusting the phase of the sampling clock that defines a specific sampling period.

  The decoding unit 20 detects the decoding result candidate of the decoding target signal input from the sampling circuit 18 by the maximum likelihood method, detects the maximum likelihood decoding result, and corrects the error in the detected decoding result to correct the decoding target. Decode the signal. In general, the maximum likelihood method is also called a maximum likelihood decoding method.

  The decoding unit 20 includes a Viterbi detector 28 and an error correction circuit 30. The Viterbi detector 28, which is an example of a detection unit according to the present invention, detects a decoding result candidate of a decoding target signal input from the sampling circuit 18 by a maximum likelihood method using a Viterbi algorithm and detects the maximum likelihood decoding result. To do.

The Viterbi detector 28 selects the maximum likelihood path from the detection paths which are a plurality of paths branched by connecting the basic paths one after another (decoding result most likely). ) Is detected. The basic path refers to a sequence of expected decoding results, that is, a stylized path connecting the expected decoding results. The basic path is determined in different combinations for each provisional decoding target signal that is an ideal decoding target signal. The combination is, for example, expected decoding result base path from A ₁ to predicted decoding result B _1, base path from the expected decoding result A ₁ to predicted decoding result B _2, expected decoding result from A ₂ to expected decoding result B ₁ base path, and it refers to a collection of such basic path to the expected decoding result B ₃ from the expected results a _3.

  In this embodiment, since the combination of the basic paths of the EPR4 (Extended class-4 Partial Response) transmission path is employed, the signal level of the provisional decoding target signal is any of -2, -1, 0, +1, +2. It is. The signal level of the provisional decoding target signal is derived according to a definition table (not shown) in which a correspondence relationship between the signal level of the decoding target signal and the signal level of the provisional decoding target signal is defined. That is, when the decoding target signal is input, the Viterbi detector 28 generates a temporary decoding target signal corresponding to the input decoding target signal according to the definition table. The Viterbi detector 28 selects a combination of basic paths corresponding to the generated provisional decoding target signal, and constructs a detection path using the selected basic paths.

  The maximum likelihood path is a path having the smallest path metric among a plurality of paths included in the detection path. Path metrics refer to a value obtained by squaring the difference between the signal level of the decoding target signal and the signal level of the provisional decoding target signal. Path metrics are an example of soft information (an index indicating the probability of determination) generated based on a decoding target signal, and are calculated for each of a plurality of paths.

  The Viterbi detector 28 selects the most likely path when the maximum likelihood path metric that is the path metric of the most likely path and the next-point path metric that is the path metric of the next most likely path are in conflict. Without waiting for the input of the decoding target signal. The Viterbi detector 28 further generates a temporary decoding target signal corresponding to the input decoding target signal, and connects the basic path corresponding to the generated temporary decoding target signal to the existing detection path to thereby detect the detection path. Continue construction to update the detection path and postpone detection of the decoding result.

  The Viterbi detector 28 outputs a Viterbi detection signal (for example, a binarized signal such as “001” or “011”) indicating the decoding result determined by selecting the maximum likelihood path to the error correction circuit 30. To do. In addition, the Viterbi detector 28 outputs the decoding target signal, the provisional decoding target signal, and the path metrics input from the sampling circuit 18 to the adjustment value calculation circuit 24.

  An error correction circuit 30 which is an example of a correction unit according to the present invention performs final decoding by correcting an error in the Viterbi detection signal input from the Viterbi detector 28, and outputs a final decoded signal indicating the final decoding result. Output to the computer 22.

  As an example, as shown in FIG. 13, the magnetic information reproducing apparatus 200 according to the first conventional technique has an adjustment value calculation circuit 202. The adjustment value calculation circuit 202 calculates a timing adjustment value based on the comparison result between the output of the sampling circuit 18 and the threshold value. Here, the timing adjustment value refers to an adjustment value used to adjust the timing at which the decoding target signal is sampled. The sampling circuit 18 adjusts the sampling timing according to the timing adjustment value calculated by the adjustment value calculation circuit 202, and samples the digital signal at the adjusted timing. However, since the reliability of the output of the sampling circuit 18 is lower than the reliability of the output of the decoding unit 20, it is difficult to obtain an accurate decoding result in a system having an SNR of, for example, 12 decibels or less.

  As an example, as shown in FIG. 14, the magnetic information reproducing apparatus 300 according to the second prior art has an adjustment value calculation circuit 302. The adjustment value calculation circuit 302 calculates a timing adjustment value based on the final decoding result of the decoding unit 20, that is, the output of the error correction circuit 30. The sampling circuit 18 adjusts the sampling timing according to the timing adjustment value calculated by the adjustment value calculation circuit 302, and samples the digital signal at the adjusted timing. However, since a delay of several hundred bits occurs until the final decoding result regarding the decoding target signal sampled by the sampling circuit 18 is obtained, it becomes difficult to follow relatively fast jitter, and reproduction of magnetic information fails. There is a fear. Note that jitter refers to temporal shift or fluctuation of a signal due to, for example, stick-slip or dropout.

  Therefore, in the magnetic information reproducing apparatus 10, the adjustment value calculation circuit 24 calculates the likelihood of the decoding result candidate estimated by the Viterbi detector 28 based on the path metrics input from the Viterbi detector 28. A timing adjustment value is calculated using the likelihood. Then, the adjustment value calculation circuit 24 outputs the calculated timing adjustment value to the sampling circuit 18, and the sampling circuit 18 adjusts the sampling timing according to the timing adjustment value, and samples the digital signal at the adjusted timing. The likelihood is calculated using soft information generated based on the decoding target signal.

  Next, the operation of the magnetic information reproducing apparatus 10 will be described. In the following, for convenience of explanation, a case where a digital signal for magnetic information recorded on the magnetic tape 26 is input from the equalizer 16 to the sampling circuit 18 at a specific period will be described.

  First, a sampling process executed by the sampling circuit 18 during a period in which a digital signal is input from the equalizer 16 to the sampling circuit 18 will be described with reference to FIG.

In the sampling process shown in FIG. 2, first, in step 100, the sampling circuit 18 waits until the sampling time comes. In step 100, when the sampling time comes, the process proceeds to step 102. Note that the sampling timing is adjusted according to the timing adjustment value τ _{k + 1} obtained by executing the sampling timing adjustment process described later by the adjustment value calculation circuit 24.

  In step 102, the sampling circuit 18 samples the decoding target signal and outputs it to the Viterbi detector 28, and then proceeds to step 104.

  In step 104, the sampling circuit 18 determines whether or not a condition for ending this sampling process is satisfied. As an example of the condition for ending the sampling process, there is a condition that an instruction signal for instructing to stop the operation of the magnetic information reproducing apparatus 10 is input to the sampling circuit 18. If it is determined in step 104 that the conditions for ending this sampling process are not satisfied, the determination is negative and the routine proceeds to step 100. In step 104, when the condition for ending the main sampling process is satisfied, the determination is affirmed and the main sampling process is ended.

  Next, a Viterbi detection process executed by the Viterbi detector 28 when a decoding target signal is input from the sampling circuit 18 will be described with reference to FIG.

  In the Viterbi detection process shown in FIG. 3, first, in Step 110, the Viterbi detector 28 stands by until a decoding target signal used for Viterbi detection is input from the sampling circuit 18. In step 110, when a decoding target signal is input, the process proceeds to step 112.

  In step 112, the Viterbi detector 28 generates a provisional decoding target signal corresponding to the decoding target signal input in step 110, and then proceeds to step 114.

  In step 114, the Viterbi detector 28 selects a basic path combination corresponding to the provisional decoding target signal generated in step 114 from the basic path combinations of the EPR4 transmission path, and detects using the selected basic path combination. A path is constructed, and then the process proceeds to step 116.

  In step 116, the Viterbi detector 28 calculates path metrics for each of a plurality of paths included in the detection path constructed in step 114, and then proceeds to step 118.

  In step 118, the Viterbi detector 28 refers to the path metrics calculated in step 116 to determine whether there is one maximum likelihood path from the currently generated paths. In step 118, if there is one path with the maximum likelihood from the currently generated paths, the determination is affirmed and the routine proceeds to step 124. If it is determined in step 118 that there is no maximum likelihood path from the currently generated paths, the determination is negative and the routine proceeds to step 120.

  In step 120, the Viterbi detector 28 determines whether or not the elapsed time from the start of sampling by the sampling circuit 18 to the current time exceeds an allowable upper limit time. Here, the allowable upper limit time refers to an upper limit time in which the sampling circuit 18 is allowed to shift in sampling timing. As an example of the upper limit time, a predetermined time can be cited as a time during which jitter generated with the conveyance of the magnetic tape 26 is allowable. In the present embodiment, 6 bits are employed as the delay bits corresponding to the upper limit time. Therefore, in this step 120, whether or not the number of delay bits after sampling is started by the sampling circuit 18 exceeds 6 bits. Is determined.

  In step 120, if the elapsed time from the start of sampling in the sampling circuit 18 to the current time does not exceed the allowable upper limit time, the determination is negative and the routine proceeds to step 110. In step 120, if the elapsed time from the start of sampling in the sampling circuit 18 to the current time exceeds the allowable upper limit time, the determination is affirmed and the routine proceeds to step 122.

  In step 122, the Viterbi detector 28 adjusts the decoding target signal input in step 110, the provisional decoding target signal generated in step 112, and the path metrics of each of a plurality of paths included in the detection path. 24, and then proceeds to step 110.

  In step 124, the Viterbi detector 28 outputs a Viterbi detection signal indicating the maximum likelihood decoding result determined by selecting one maximum likelihood path from the detection paths to the error correction circuit 30, and then in step 126. Migrate to The error correction circuit 30 performs error correction on the Viterbi detection signal input from the Viterbi detector 28 and outputs the signal obtained by the correction to the computer 22 as a final decoded signal.

  In step 126, the Viterbi detector 28 determines whether or not a condition for ending this Viterbi detection process is satisfied. As an example of the condition for ending the Viterbi detection process, there is a condition that an instruction signal instructing to stop the operation of the magnetic information reproducing apparatus 10 is input to the decoding unit 20.

  If the condition for ending the Viterbi detection process is not satisfied in step 126, the determination is negative and the routine proceeds to step 110. In step 126, when the condition for ending this Viterbi detection process is satisfied, the determination is affirmed and the Viterbi detection process is ended.

  Next, the sampling timing adjustment process executed by the adjustment value calculation circuit 24 will be described with reference to FIG.

  In the sampling timing adjustment process shown in FIG. 4, first, in step 130, the adjustment value calculation circuit 24 waits until a decoding target signal, a provisional decoding target signal, and path metrics are input from the Viterbi detector 28. When the decoding target signal, the provisional decoding target signal, and the path metrics are input from the Viterbi detector 28 in step 130, the process proceeds to step 132.

In step 132, the adjustment value calculation circuit 24 uses the following formulas (1), (2), and (3) to calculate the timing adjustment value τ _{k + 1} (an example of the extraction timing adjustment amount according to the present invention). After that, the process proceeds to step 134.

ε _k ≡y _k d _k−1 −y _k−1 d _k (1)
θ _k = θ _k−1 + g _k βε _k (2)
τ _{k + 1} = τ _k + g _k αε _k + θ _k (3)

  In Equations (1), (2), and (3), k is a natural number, and ε is a phase error between the decoding target signal and the provisional decoding target signal, that is, to the Viterbi detector 28. The phase error between the input and the output from the Viterbi detector 28 to the adjustment value calculation circuit 24, y is the signal level of the decoding target signal, d is the signal level of the provisional decoding target signal, and τ is The timing adjustment value, g is the probability that the Viterbi detector 28 correctly detects the decoding result (in other words, the probability that the Viterbi detector 28 makes the correct determination), and θ is the timing of sampling by the sampling circuit 18. (Frequency error (phase error differential term)) and α and β are adjustment terms (fixed values).

In addition, the probability g _k in Equation (2) and Equation (3) is a probability that defines the likelihood of the decoding result estimated by the decoding unit 20, and is defined by a function represented by the following Equation (4). Yes.

In Equation (4), M ¹ is the maximum likelihood path metric, and M ² is the next path metric. This means that the probability g _k expressed by the equation (4) varies according to soft information obtained from separate paths of maximum likelihood path metrics and next-point path metrics, and the larger the value of the probability g _k is, the larger the probability g _k is. It means that the likelihood is high. Here, the case where the probability is calculated using the maximum likelihood path metric and the next-point path metric is illustrated, but the present invention is not limited to this. For example, the probability may be calculated that further considers the path metrics of the likely paths after the next point, and the competitive path metrics that compete with the maximum likelihood path metrics in selecting the maximum likelihood path, and the maximum The probability may be calculated using the likelihood path metrics.

Further, the probability g _k in the mathematical formulas (2) and (3) may be defined by the following mathematical formula (5).

Further, the probability g _k in the formulas (2) and (3) may be defined by the following formula (6). In Equation (4) and Equation (5), only the next-point path metric M ² is considered as the path metric of a specific conflict path, whereas in Equation (6), all the competing paths M _i are considered. ing.

  In step 134, the adjustment value calculation circuit 24 outputs the timing adjustment value calculated in step 132 to the sampling circuit 18, and then ends the sampling timing adjustment process.

  In this way, when the timing adjustment value is output to the sampling circuit 18, the sampling circuit 18 adjusts the sampling timing according to the set timing adjustment value, and samples the digital signal at the adjusted sampling timing.

Note that the adjustment of the sampling time is realized by adjusting the result of nonlinear interpolation such as Nyquist interpolation with the time adjustment value, but the present invention is not limited to this. The adjustment of the sampling time may be realized, for example, by adjusting the result of linear interpolation between digital signals whose input times from the equalizer 16 are adjacent to each other by the time adjustment value. In this case, for example, when a digital signal is input to the sampling circuit 18 at intervals of 10 nanoseconds and the timing adjustment value is 5 nanoseconds, sampling is performed at a time corresponding to the center point of the digital signal _xk and the digital signal _{xk + 1.} The sampling time is adjusted to be performed.

Here, as an example, as shown in FIG. 5, the probability g _k tends to decrease at an error occurrence location (in the example shown in FIG. 5, an error flag is set). Therefore, in the magnetic information reproducing apparatus 10, the sampling timing is adjusted according to the timing adjustment value τ _{k + 1} that varies according to the probability g _k , so that even if the reading result of the magnetic head 12 changes from moment to moment, The signal to be decoded following the change is sampled.

  FIGS. 6 to 10 show the timing of sampling in a system with a relatively low SNR of about 12 decibels to 12.5 decibels by delaying by an allowable upper limit of 6 bits from the viewpoint of tracking the jitter. A comparative example when adjusted is shown.

FIG. 6 shows a case where the total number of error bits (broken line) with respect to the processing data length when the probability g _{k represented} by the equation (4) is used, and the probability g _k is a fixed value (here, “1” as an example). An example of the total number of error bits (solid line) for the processing data length is shown. In the example shown in FIG. 6, the total number of error bits with respect to the processing data length when the probability g _k is a fixed value increases rapidly at a certain processing data length, whereas the probability g expressed by Equation (4) _The total number of error bits for the processing data length when _k is used does not increase rapidly.

7 and 9, the SNR (broken line) with respect to the processing data length when the probability g _{k represented} by the equation (4) is used, and the probability g _k are fixed values (here, “1” as an example). An example of the SNR (solid line) with respect to the processing data length is shown. In the example shown in FIG. 7, the SNR with respect to the processing data length when the probability g _k is a fixed value is drastically decreased at a certain processing data length, whereas the probability g _k expressed by the equation (4) is The SNR for the processing data length when used is not drastically reduced. Further, in the example shown in FIG. 9, the SNR for the processing data length when the probability g _k is a fixed value fluctuates for a plurality of times, whereas the probability g _k shown by the equation (4) is used. In this case, the SNR with respect to the processing data length is stable.

8 and 10, the phase error (broken line) with respect to the processing data length when the probability g _{k represented} by the equation (4) is used, and the probability g _k are fixed values (here, “1” as an example). An example of the phase error (solid line) with respect to the processing data length is shown. In the example shown in FIG. 8, the phase error with respect to the processing data length when the probability g _k is a fixed value is suddenly increased and fixed at a certain processing data length, whereas it is expressed by Equation (4). The phase error with respect to the processing data length when the probability _gk to be used is not increased rapidly. Further, in the example shown in FIG. 10, the phase error with respect to the processing data length when the probability g _k is a fixed value fluctuates for a plurality of times, whereas the probability g _k shown by the equation (4) is used. In this case, the phase error with respect to the processing data length is stable.

As described above, when the probability g _k is set to a fixed value, when jitter occurs, the sampling operation by the sampling circuit 18 is not stabilized, the sampling timing is shifted, and the shift is fixed. On the other hand, when the probability g _k shown by the equation (4) is used, even if jitter occurs, the sampling time lag is immediately corrected. This uses the probability g _k that varies from moment to moment according to the maximum likelihood path metric and the next-point path metric, and the reliability of the timing adjustment value when the probability g _k shown by Equation (4) is used. This is because the reliability of the timing adjustment value when the probability g _k is a fixed value is higher.

  As described above, in the magnetic information reproducing apparatus 10, the decoding unit 20 estimates the decoding result candidate of the decoding target signal by the maximum likelihood method and detects the maximum likelihood decoding result. The adjustment value calculation circuit 24 adjusts the sampling timing of the decoding target signal using the likelihood of the decoding result candidate estimated by the decoding unit 20. Therefore, the magnetic information reproducing apparatus 10 achieves both high-precision sampling of a specific decoding target signal and suppression of a delay caused by adjustment of the sampling timing, as compared with the first conventional technique and the second conventional technique. be able to.

  Further, in the magnetic information reproducing apparatus 10, the decoding unit 20 includes a Viterbi detector 28 and an error correction circuit 30, and sampling timing using the likelihood of candidate decoding results estimated by the Viterbi detector 28. Is adjusted. Therefore, the magnetic information reproducing apparatus 10 can suppress a delay that occurs due to the adjustment of the sampling timing, as compared with the case where the sampling timing is adjusted based on the output of the error correction circuit 30.

  Further, in the magnetic information reproducing apparatus 10, the sampling timing is adjusted on the condition that a predetermined time has passed as the upper limit time in which the sampling timing deviation is allowed after the sampling of the decoding target signal is started. Be started. Therefore, the magnetic information reproducing apparatus 10 adjusts the sampling timing before the predetermined time has elapsed as the upper limit time in which the deviation of the sampling timing is permitted after the sampling of the decoding target signal is started. Compared to the above, the reliability of the sampling time can be improved.

  Further, in the magnetic information reproducing apparatus 10, the sampling is performed every time the signal to be decoded is sampled after a predetermined time has passed as the upper limit time in which the deviation of the sampling timing is allowed after the sampling is started. Likelihood based on the decoding target signal is calculated. Each time the likelihood is calculated, the sampling timing is adjusted using the calculated likelihood. Therefore, the magnetic information reproducing apparatus 10 uses the likelihood calculated first after a predetermined time has passed as the upper limit time in which the sampling time deviation is allowed after the sampling is started. Compared with the case where the adjustment is performed only once, even when the decoding target signal fluctuates, it is possible to ensure high reliability with respect to the sampling timing.

  Further, in the magnetic information reproducing apparatus 10, every time the decoding target signal is sampled, the likelihood based on the sampled decoding target signal is calculated regardless of whether or not the decoding unit 20 detects the maximum likelihood decoding result. Is done. Each time the likelihood is calculated, the sampling timing is adjusted using the calculated likelihood. Therefore, the magnetic information reproducing apparatus 10 suppresses the delay caused by the adjustment of the sampling time, compared with the case where the sampling time is adjusted after waiting for the decoding unit 20 to detect the maximum likelihood decoding result. can do.

Further, in the magnetic information reproducing apparatus 10, the amount of adjustment at the sampling time is reduced as the likelihood calculated by the adjustment value calculation circuit 24 is lower. As a result, it is possible to suppress the occurrence of an undesired adjustment amount due to an erroneous determination by the Viterbi detector 28. Therefore, the magnetic information reproducing apparatus 10 can increase the adjustment accuracy of the sampling time as compared with the case where the adjustment amount of the sampling time is determined regardless of the likelihood. Note that the higher the probability g _k , the higher the likelihood.

  Moreover, in the magnetic information reproducing | regenerating apparatus 10, likelihood is calculated using the soft information produced | generated based on the decoding object signal. Therefore, the magnetic information reproducing apparatus 10 can obtain a highly accurate likelihood as compared with the case where the likelihood is defined without using soft information.

  Further, in the magnetic information reproducing apparatus 10, the likelihood is defined by the probability that the decoding result is correctly detected by the decoding unit 20, and the probability that defines the likelihood is calculated using the maximum likelihood path metrics and the competitive path metrics. Probability is adopted. Therefore, the magnetic information reproducing apparatus 10 can obtain a highly accurate likelihood as compared with the case where the likelihood is defined only by the maximum likelihood path metric or the competitive path metric.

  Moreover, in the magnetic information reproducing | regenerating apparatus 10, the sampling time is adjusted according to the time adjustment value calculated using Numerical formula (1)-(4). Therefore, the magnetic information reproducing apparatus 10 can adjust the sampling timing with high accuracy compared to the case where the sampling timing is adjusted only by the phase error calculated using Expression (1).

  Further, in the magnetic information reproducing apparatus 10, the maximum likelihood method based on the Viterbi algorithm is adopted as the maximum likelihood method used for estimation of decoding result candidates. Therefore, the magnetic information reproducing apparatus 10 can obtain a high cost-effectiveness as compared with the case where the maximum likelihood method using an algorithm other than the Viterbi algorithm is adopted.

  In the magnetic information reproducing apparatus 10, magnetic information is read from the magnetic tape 26. Therefore, the magnetic information reproducing apparatus 10 has a high level of a specific decoding target signal even when a change in the conveyance speed of the magnetic tape 26 occurs locally as compared with the first conventional technique and the second conventional technique. It is possible to achieve both accurate sampling and suppression of delay caused by adjustment of the sampling timing.

  In the above embodiment, the magnetic information reproducing apparatus 10 having the Viterbi detector 28 to which the decoding target signal is directly input from the sampling circuit 18 is illustrated, but the present invention is not limited to this. For example, as shown in FIG. 11, a magnetic information reproducing apparatus 50 in which an amplitude adjusting circuit 19 is interposed between the sampling circuit 18 and the Viterbi detector 28 may be employed.

  An example of the amplitude adjustment circuit 19 is an AGC (automatic gain controller). In the amplitude adjustment circuit 19, the amplitude component of the decoding target signal output from the sampling circuit 18 is adjusted. Then, the Viterbi detector 28 estimates the decoding result candidate of the decoding target signal whose amplitude component is adjusted by the amplitude adjusting circuit 19 by the maximum likelihood method, and detects the maximum likelihood decoding result. Therefore, the magnetic information reproducing apparatus 50 can improve the detection accuracy of the maximum likelihood decoding result by the Viterbi detector 28 as compared with the case where the decoding target signal is directly input from the sampling circuit 18 to the Viterbi detector 28.

  In the above embodiment, the case where the sampling timing is adjusted using the likelihood obtained from the Viterbi detector 28 is exemplified. However, the error correction result estimated when the error is corrected by the maximum likelihood method. The sampling timing may be adjusted using the likelihood of the candidate.

  As an example, as shown in FIG. 12, the magnetic information reproducing device 70 has an adjustment value calculating circuit 60 instead of the adjustment value calculating circuit 24 as compared with the magnetic information reproducing device 10, and a decoding unit instead of the decoding unit 20. The difference is that it has 62. The decoding unit 62 differs from the decoding unit 20 in that it has a Viterbi detector 63 instead of the Viterbi detector 28 and an error correction circuit 64 instead of the error correction circuit 30.

  The Viterbi detector 63 outputs the Viterbi detection signal to the error correction circuit 64 without outputting the decoding target signal, provisional decoding target signal, and path metrics to the adjustment value calculation circuit 60.

  The error correction circuit 64 detects the error correction result candidate of the decoding result detected by the Viterbi detector 63 by using the maximum likelihood method, and detects the maximum likelihood error correction result, so that the Viterbi detector 63 detects the error correction result. An error in the decoded result is corrected.

  The error correction circuit 64 outputs the Viterbi detection signal, the temporary Viterbi detection signal that is an ideal Viterbi detection signal, and the path metrics constructed by the error correction circuit 64 to the adjustment value calculation circuit 60. The signal level of the Viterbi detection signal corresponds to the signal level of the decoding target signal used for calculating the timing adjustment value in the above embodiment. Further, the signal level of the provisional Viterbi detection signal corresponds to the signal level of the provisional decoding target signal used for calculating the timing adjustment value in the above embodiment.

  The adjustment value calculation circuit 60 calculates a timing adjustment value from the Viterbi detection signal, the provisional Viterbi detection signal, and the path metrics input from the error correction circuit 64 using Equations (1) to (4), and calculates the calculated timing adjustment. The value is output to the sampling circuit 18.

  The sampling circuit 18 adjusts the sampling timing according to the timing adjustment value input from the adjustment value calculation circuit 60, and samples the decoding target signal at the adjusted timing.

  The sampling adjustment by the sampling circuit 18 and the adjustment value calculation circuit 60 is started on the condition that the elapsed time from the start of sampling in the sampling circuit 18 to the current upper limit exceeds an allowable upper limit time. In other words, the adjustment value calculation circuit 60 determines the error correction result candidate every time the Viterbi detector 63 detects a decoding result after an allowable upper limit time has elapsed since the sampling circuit 18 started sampling. The likelihood based on the decoding result detected by the Viterbi detector 63 is calculated as the likelihood. The adjustment value calculation circuit 60 calculates a timing adjustment value using the calculated likelihood and outputs it to the sampling circuit 18 every time the likelihood is given. As an example of the upper limit time, a predetermined time can be cited as a time during which jitter generated with the conveyance of the magnetic tape 26 is allowable.

  As described above, in the magnetic information reproducing apparatus 70, the error correction circuit 64 estimates the error correction result candidate of the decoding result by the maximum likelihood method and detects the maximum likelihood error correction result. Resulting errors are corrected. The adjustment value calculation circuit 60 adjusts the sampling timing of the decoding target signal using the likelihood of the error correction result candidate estimated by the error correction circuit 64. Therefore, the magnetic information reproducing apparatus 70 achieves both high-precision sampling of a specific decoding target signal and suppression of a delay caused by adjustment of the sampling timing, as compared with the first and second conventional techniques. be able to.

  Further, an error correction circuit (post error correction circuit) having a function of estimating the error correction result by the maximum likelihood method may be provided after the error correction circuit 30, and the soft information obtained by the post error correction circuit may be provided. The likelihood may be calculated using.

  Further, in the above embodiment, the likelihood based on the sampled decoding target signal is calculated every time sampling is performed after a predetermined time has elapsed since the sampling was started. However, the present invention is not limited to this. For example, the likelihood based on the sampled decoding target signal may be calculated every time sampling is performed before a predetermined time as an allowable upper limit time has elapsed since the sampling was started. Good. In this case, each time the likelihood is calculated, the sampling timing is adjusted using the calculated likelihood. As a result, the magnetic information reproducing apparatus 10 suppresses a decrease in the adjustment accuracy of the sampling timing due to fluctuations in the decoding target signal, compared to the case where the sampling timing adjustment using the likelihood is performed only once. can do.

  In the above embodiment, the EPR4 transmission line is exemplified, but the present invention is not limited to this, and other transmission lines such as a PR4 (Partial Response class-4) transmission line and a GPR (Generalized Partial Response) are used. May be adopted.

  In the above embodiment, the Viterbi algorithm has been exemplified. However, the present invention is not limited to this, and other maximum likelihood decoding such as an FDTS (Fixed-depth tree search) algorithm, a Fano algorithm, a stack algorithm, etc. An algorithm may be adopted.

  In the above embodiment, the sampling circuit 18 and the adjustment value calculation circuit 24 are separated, but the present invention is not limited to this, and the sampling circuit 18 and the adjustment value calculation circuit 24 are integrated. It may be made.

  Moreover, although the case where magnetic information was read from the magnetic tape 26 was illustrated in the said embodiment, this invention is not limited to this. For example, the present invention can be applied to a case where magnetic information is read from a magnetic disk.

  Further, in the above embodiment, the case where the sampling circuit 18, the decoding unit 20, and the adjustment value calculation circuit 24 adjust the sampling timing of the decoding target signal is illustrated, but the present invention is not limited to this. Alternatively, the existing technology and the present invention may be combined.

  For example, by feeding back the output of the decoder to the input stage of the phase adjustment circuit, the phase of the signal input to the input stage of the phase adjustment circuit is adjusted, and an independent phase adjustment circuit is provided for each possible decoding pattern. In comparison with the existing technique (hereinafter referred to as the existing technique A) that enables optimal feedback according to the decoded pattern, the sampling circuit 18, the decoding unit 20, and the adjustment value calculation circuit 24 are implemented. May incorporate a main function (hereinafter referred to as “main function according to the present embodiment”) that adjusts the sampling timing of the decoding target signal. Further, even in a system in which a feedback delay to the input stage of the phase adjustment circuit exists, an existing technique (hereinafter referred to as the existing technique B) that minimizes the feedback delay by optimizing the gain parameter of the phase adjustment circuit. The main functions according to the present embodiment may be incorporated. In addition, since multiple channels are affected by speed fluctuations simultaneously in a tape system, not only single-channel signals but also information on speed fluctuations obtained from playback signals of other channels can be used to reduce jitter tolerance. The function of the main part according to the present embodiment may be incorporated into an existing technique for improving (hereinafter referred to as existing technique C). In addition, the main function according to the present embodiment may be incorporated into a combination of the existing technology A, the existing technology B, and the existing technology C, or the existing technology A, the existing technology B, and the existing technology C The main functions according to the present embodiment may be incorporated into the two combinations. As described above, by combining the present invention and the existing technology, a further higher effect can be expected.

In the above embodiment, the method of calculating the probability g _k by the equation (6) is exemplified, but the present invention is not limited to this. The probability g _k can be an index of the probability of decoding determination, but an example of a method for more strictly calculating the probability of correct determination will be shown below.

First, the probability g _{k defined} by Equation (6) is set as the probability p _k . Next, the probability of each bit defined by Equations (4) to (6) using a known signal in advance (a signal obtained from another determination method capable of determining whether or not a correct determination has been made). By associating the value of p _k with the correctness / incorrectness of the determination at that bit, it is possible to statistically calculate the probability g _k of performing a correct determination at that bit when the value of each probability p _k is obtained. It becomes. By confirming the correspondence between the probability p _k and the probability g _k by a pre-calibration process or the like and creating a conversion table in advance, the probability p _k is not a mere likelihood index but a statistical probability. It is possible to calculate the probability g _k from the above. As a method for calculating the probability p _k, for example, a method using Equation (4) to (6), and a method described below.

In the above embodiment, the value of the path metric (maximum likelihood path metric) of the favorite path and the path metric (competing path metric) of the competing path is used to calculate the probability g _k that the decoding result is correctly detected. The invention is not limited to this, and the probability can be determined more precisely.

As described above, the probability g _k described in the above embodiment is one of the indices indicating likelihood (= probability p _k ). In the example shown in the equation (6), the favorite path is all other values. It expresses the certainty for the competing path. On the other hand, among the competing paths at this time, although the latest bit is different from the favorite path, there is a path that converges to the same path after a certain delay. In addition, although it is a different path, there may be a case where the bit determination (“0” or “1”) when going back by a certain delay is the same as the main path as a result. What actually matters is that (i) at the time of a predetermined delay bit (past), the determination at that time is either “0” or “1”, (ii) What is the probability that the determination of “0” or “1” is correct?

  In the above embodiment, the following methods are taken into consideration with respect to the above (i) and (ii).

For (i) above, when the allowable time (bit) has elapsed, the path with the smallest path metric is selected as the correct path, and the timing of the timing is determined using the past decoding results for the delay bits of that path. Make adjustments. For (ii) above, the probability g _k is calculated using the path metric values of the correct path candidate selected above and the other competing paths.

  On the other hand, in order to correctly estimate the above (i) and (ii), the following method is effective.

With respect to all paths including the favorite path and the competing path, the decoding result of “0” and “1” at the past time and the path metrics of each path for a predetermined bit (delay bit) in each path. Using the value of (for example, current path metrics), determination is made according to the following formulas (7) and (8), and the probability (probability) p _k is calculated. That is, the comparison is performed not by comparing the favorite path and the competing path but by comparing the path with the decoding result “0” and the path “1” at the past specified bit.

In Equations (7) and (8), “a” is a decoding result in the past for a predetermined delay bit of each path, and “A” is finally based on these pieces of information. It is the decoding result to be determined. Here, the probability (probability) of A = 0 and A = 1 is set to P _k ^{A = 0} and P _k ^{A = 1} , respectively, and the path metrics of each path where a = 0 and a = 1 are set to M ^a , respectively. ^{= 0} and M ^{a = 0} .

In addition, according to the magnitude relationship of P _k ^{A = 0} and P _k ^{A = 1,} a provisional determination result used for detecting the phase error is obtained according to the following inequality (9) and inequality (10).

  As described above, determination and likelihood calculation are performed in consideration of the path metrics of all paths including the favorite path and the competing path and the decoding results of “0” and “1” in each path. As a result, the accuracy of determination can be improved and more accurate likelihood calculation can be performed as compared with the method of determining based on the result of only the favorite path.

  In the above embodiment, the decoding result of “0” and “1”, and the case of using the Viterbi algorithm and the value of the path metric used for calculation in the Viterbi algorithm in order to obtain the likelihood of the decoding result. The invention is not limited to this. For example, a BCJR (Bahl, Cocke, Jelinek, Raviv) algorithm may be used as means for obtaining soft information. The BCJR algorithm may be a MAP algorithm or a sum-product algorithm (“Bahl, J. Cocke, F. Jelinek, and J. Raviv,“ Optimal Decoding of Linear Codes for the Emerging Symbols ”). IT-20 (2), pp. 284-287, March 1974. ”or“ http://www.ieice-hbkb.org/files/01/01gun — 02hen — 05.pdf ”).

  Whereas the Viterbi algorithm is aimed at selecting the maximum likelihood path (path metrics are the index for it), the BCRJ algorithm is “0” for each bit calculated from the path metrics of all the paths existing at a certain time. The purpose is to quantify the likelihood of "" or "1". For this reason, in the Viterbi algorithm, only one path is selected when two competing paths intersect. At that time, information (path metrics) of the unselected path is lost, but in the BCJR algorithm, basically The path metric values for all paths are stored.

  Since the normal BCJR algorithm uses all data for one sector (usually data of about 8000 bits) when calculating the likelihood, there is a restriction that processing cannot be started until data for one sector is ready. Yes. For this reason, although the immediacy of determination important in the present invention and superiority with respect to the buffer memory are lacking, the sliding-window BCJR algorithm (“BENEDETTO, S., et al. Soft-output decoding algorithms in iteratives) covering this point is covered. decoding of turbo codes. See TDA Progress Report 42, 1996, 124: 63-87. "or" http://tmo.jpl.nasa.gov/progress_report/42-124G.pdf "and SoftV. ) Algorithm (“HAGENAUER, Joachim; HOEHER, Pete .. A Viterbi algorithm with soft-decision outputs and its applications In:. Global Telecommunications Conference and Exhibition'Communications Technology for the 1990s and Beyond '(GLOBECOM), 1989. IEEE IEEE, 1989. P. 1680-1686 ",". http://www.ce.lehigh.edu/˜jingli/teach/S2003turbo/notes/SOVA.pdf ”or“ http://www.ieice-hbkb.org/files/01/01gun ” 02hen_05.pdf ", or," http://www.ieice-hbkb.org/files/01/01gun_02hen_05.pdf "reference) have been developed.

  The sliding-window BCJR algorithm is a simplified version of the BCJR algorithm performed within a specified delay bit window, compared to a normal BCJR algorithm, and it is necessary to store only a predetermined delay bit value of a path metric. The likelihood of each bit is calculated with higher accuracy than the Viterbi algorithm.

  The SOVA algorithm is a development system of the Viterbi algorithm, and saves the information of the difference (= likelihood) of the path metrics of both paths when two competing paths intersect with each other. It is comprised so that it can utilize for calculation of likelihood. Compared to the BCJR algorithm and the sliding-window BCJR algorithm, which uses the latest input bit information (path metrics) and reflects the likelihood of the past bits retroactively, the SOVA algorithm is different from the normal Viterbi algorithm. Similarly, the input data is processed sequentially (the likelihood of the past bit once determined is not updated based on the subsequent input data), so the load on the circuit implementation is based on the sliding-window BCJR algorithm Has the advantage of being low.

In calculating the probability P _{k of} each bit, it is possible to calculate the likelihood with higher accuracy by using the sliding-window BCJR algorithm and the SOVA algorithm, and it is possible to calculate the gain (probability of correct determination) at the time of phase adjustment. ) A more suitable value can be selected as _gk .

10, 50, 70 Magnetic information reproducing device 12 Magnetic head 14 A / D converter 16 Equalizer 18 Sampling circuit 19 Amplitude adjustment circuit 20, 62 Decoding unit 24, 60 Adjustment value calculation circuit 28, 63 Viterbi detectors 30, 64 Error correction circuit

Claims (15)

An extraction unit for extracting the decoding target signal from the input digital signal at an extraction timing determined as a timing for extracting the decoding target signal;
A decoding unit that decodes the decoding target signal by estimating a decoding result candidate of the decoding target signal extracted by the extraction unit by a maximum likelihood method and detecting a maximum likelihood decoding result;
An adjustment unit that adjusts the extraction timing so that the amount of adjustment of the extraction timing decreases as the likelihood decreases , using the likelihood of the candidate decoding result estimated by the decoding unit;
Including a signal processing apparatus.   The decoding unit includes a detection unit that estimates the candidate by the maximum likelihood method and detects a maximum likelihood decoding result, and a correction unit that corrects an error in the maximum likelihood decoding result detected by the detection unit, The signal processing apparatus according to claim 1.   The adjustment of the extraction timing by the adjustment unit is performed on the condition that a predetermined time has passed as an upper limit time during which the deviation of the extraction timing is allowed after the extraction unit starts extracting the decoding target signal. The signal processing apparatus according to claim 1 or 2, which is started.   The adjustment unit is extracted each time the decoding target signal is extracted by the extraction unit after the predetermined time has elapsed since the extraction unit started extracting the decoding target signal. The signal processing apparatus according to claim 3, wherein the likelihood based on the decoding target signal is calculated, and the extraction timing is adjusted using the calculated likelihood every time the likelihood is calculated.   The adjustment unit calculates the likelihood based on the extracted decoding target signal every time the extraction target signal is extracted by the extraction unit, and calculates the likelihood every time the likelihood is calculated. The signal processing apparatus according to claim 1, wherein the extraction timing is adjusted using a signal.   Regardless of whether the maximum likelihood decoding result is detected by the decoding unit, the adjusting unit is configured to extract the likelihood based on the extracted decoding target signal every time the extracting unit extracts the decoding target signal. 6. The signal processing device according to claim 4, wherein the degree of extraction is calculated, and the extraction timing is adjusted using the calculated likelihood each time the likelihood is calculated. The likelihood signal processing apparatus according to any one of claims 1 to 6 is a likelihood calculated by using the soft information generated on the basis of the decoding target signal. The likelihood is defined by a probability that a decoding result is correctly detected by the decoding unit,
The signal processing apparatus according to claim 7 , wherein the probability is a probability calculated using a maximum likelihood path metric and a competitive path metric that are the soft information. The adjustment unit sets k as a natural number, ε as a phase error between the decoding target signal and an ideal decoding target signal, y as a signal level of the decoding target signal, and d as the ideal decoding target signal. Equation (1) and Equation (1) below, where the signal level is τ, τ is the adjustment amount of the extraction timing, g is the probability, θ is the differential term of the extraction timing error, and α and β are adjustment terms. The signal processing apparatus according to claim 8 , wherein the extraction timing is adjusted using an adjustment amount τ _{k + 1} obtained by (2) and Equation (3).
ε _k ≡y _k d _k−1 −y _k−1 d _k (1)
θ _k = θ _k−1 + g _k βε _k (2)
τ _{k + 1} = τ _k + g _k αε _k + θ _k (3) The signal processing apparatus according to any one of claims 1 to 9 , wherein the maximum likelihood method is a maximum likelihood method based on a Viterbi algorithm. An amplitude component adjusting unit that adjusts an amplitude component of the decoding target signal;
The decoding unit detects a maximum likelihood decoding result by estimating a decoding result candidate of the decoding target signal whose amplitude component is adjusted by the amplitude component adjusting unit by the maximum likelihood method. The signal processing device according to any one of 0 . A read head for reading magnetic information from a magnetic recording medium;
A generator for generating a digital signal from magnetic information read by the read head;
The signal processing device according to any one of claims 1 to 11, wherein a digital signal generated by the generation unit is input;
Magnetic information reproducing apparatus including: The magnetic recording medium is a magnetic information reproducing apparatus according to claim 1 2 is a magnetic tape. The adjustment of the extraction timing by the adjustment unit included in the signal processing device can allow jitter generated by the conveyance of the magnetic tape after the extraction of the decoding target signal by the extraction unit included in the signal processing device is started. magnetic information reproducing apparatus according to claims 1 to 3, a predetermined time as a time is started on condition that it has passed. From the input digital signal, extract the decoding target signal at an extraction timing determined as a timing for extracting the decoding target signal,
The decoding target signal is decoded by estimating the decoding result candidate of the extracted decoding target signal by the maximum likelihood method and detecting the maximum likelihood decoding result,
A signal processing method including adjusting the extraction timing so that the amount of adjustment of the extraction timing decreases as the likelihood decreases , using the estimated likelihood of the candidate decoding result.