KR19990038679A - Viterbi detection method and apparatus - Google Patents

Abstract

The method and apparatus for detecting a viterbi according to the present invention determine an object to be decoded in accordance with transfer functions of various optical channel models and store the decoded object in a register and then perform viterbi detection using the stored decoding target information, And an overflow control unit is provided between the add-compare-select unit and the state evaluation value memory so as to prevent the overflow of the overflow control unit and to operate at a high speed. Since the present invention is applied to a high-speed optical storage medium, it is possible to detect an EFM-modulated signal by a simple equalizer alone, and various states and branches can be eliminated by the RLL characteristic of EFM modulation, .

Description

Viterbi detection method and apparatus

More particularly, the present invention relates to a Viterbi detection method and apparatus capable of detecting a high-speed EFM-modulated signal with a simple equalizer without requiring an accurate equalizer.

In general, the characteristics of the optical channel depend on the characteristics of the optical pickup device. The characteristics of the optical pickup are similar to those of a low-pass filter. Upon detection on a bit-by-bit basis, the magnitude of the signal level must be detected to restore the original data stream. At this time, when the optical disc rotates at a high speed, the waveform of the output signal is distorted. In a high-density recording medium such as a DVD (digital versatile disk), noise components in the read signal are increased due to crosstalk between adjacent tracks. Therefore, in order to compensate for such distortion, a high-performance digital or analog equalizer is required.

In a signal recovery apparatus for a conventional high-speed optical disk, a signal output from the optical pickup is subjected to level detection via an expensive analog equalizer, thereby restoring the original signal. The analog equalizer compensates for the waveform of the signal output from the optical pickup. The compensated analog signal is recovered to the original signal by detecting the magnitude of the compensated analog signal bit by bit in the level detector. In this case, in a high-speed optical disc player, distortion of a signal output from the optical pickup is increased, and a highly sophisticated analog equalizer is required, which leads to an increase in manufacturing cost.

Particularly, in a high-speed optical system such as a 30-times CD-ROM or a 4.5-times DVD ROM / RAM, the analog output of the optical pickup is expected to have an ideal response due to distortion due to inter- symbol interference (ISI) It is difficult to do.

SUMMARY OF THE INVENTION The present invention has been made in view of the above problems, and it is therefore an object of the present invention to provide an optical disc drive capable of increasing the operation speed even with a simple equalizer without a high-precision equalizer in reproduction of a high- And to provide a Viterbi detection method capable of detecting non-detection.

It is another object of the present invention to provide an optical disc reproducing apparatus and an optical disc reproducing method which can increase the operating speed without using a high-precision equalizer in reproducing a high-speed high-density optical disc, And to provide a Viterbi detection device capable of Viterbi detection.

1 is a block diagram of a PRML system to which the Viterbi detection apparatus of the present invention is applied.

FIG. 2 shows a Trellis diagram of the optical channel model.

3 is a circuit diagram showing an example of the overflow controller of FIG.

Figure 4 shows a diagram of the path memory of Figure 1;

5A and 5B are diagrams showing the cell structure constituting the channel memory when the branch is 2 and 1, respectively.

6A, 6B, and 6C are graphs showing the signal ratios according to the transfer function of each channel model.

7 is a circuit diagram showing an example of the branching amount calculator of FIG.

8 is a graph showing a frequency response curve of transfer functions of each channel model.

Description of the Related Art

20 ... Momentum amount calculator, 30 ... Addition-comparison-

40 ... overflow controller, 50 ... status evaluation memory

60 ... passage memory.

According to an aspect of the present invention, there is provided a Viterbi detection method including the steps of: (a) determining a channel model to be decoded; (b) receiving the read data from the optical pickup and calculating a difference absolute value between the data and each status reference value; (c) calculating the Euclidean distance by adding the two previous state evaluating quantities and the branch quantities which are transited to the current state, respectively, on the basis of the respective quantities calculated using the calculated difference absolute values, and Selecting a smaller value among the two values to determine a new state evaluation amount and generating a path selection signal; (d) preventing an overflow by using a difference between a largest value and a smallest value of the determined state estimation amount converged to a predetermined value; (e) storing the determined overflow-protected new state estimate and sending the stored output to the previous state estimate in step (c); And (f) receiving the path selection signal from the step (c) and decoding the original signal using a register exchange structure.

As another example, the present invention may cause step (d) to follow step (e).

Preferably, in the optical channel model determining step (a), a decoding target model is determined according to transfer functions of various optical channel partial response models, and the information is stored in a register.

Preferably, the overflow prevention step (d) is characterized in that ^a difference between the largest value and the smallest value of the state estimation amount is always smaller than 2 ^{a + 2} when a is the bit width of the branch quantity.

Advantageously, said overflow protection step (d) is characterized by bitwise and gating the most significant bit or other bits.

According to another aspect of the present invention, there is provided a Viterbi detection apparatus comprising: a register for determining and outputting a channel model to be decoded; A division calculation unit for receiving the read data from the optical pickup and calculating a difference absolute value between the read data and each state reference value; The Euclidean distances are obtained by adding the previous two state estimates and the branch quantities which are transited to the current state based on the respective branch quantities calculated using the calculated difference absolute values and then adding the two added values An add-compare-select unit for comparing the two values and selecting a smaller value among the two values to determine a new state evaluation amount and generating a path selection signal; An overflow control unit that uses the fact that the difference between the largest value and the smallest value of the determined state estimation amount converges to a predetermined value; A state evaluation memory storing the determined overflow-protected new state estimate and transmitting the stored output to a previous state estimate of the add-compare-select; And a path memory for receiving a path selection signal from the add-compare-selection unit and removing a predetermined number of states and branches using a register exchange structure.

In another embodiment, the present invention allows the overflow control unit to be connected between the output of the state estimation memory and the input of the add-compare-select unit.

Preferably, a model to be decoded is selected according to transfer functions of various optical channel partial response models according to the value of the register unit.

Preferably, a difference between a maximum value and a minimum value of the state estimation amount becomes smaller than 2 ^{a + 2} when the overflow control unit is a bit width of the branch quantity.

Preferably, the overflow control unit bitwise and gates the most significant bit or the other bit.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, preferred embodiments of a method and an apparatus for detecting a viterbi according to the present invention will be described in detail with reference to the accompanying drawings.

The optical channel can be modeled with simple partial response (PR) characteristics, and the Viterbi detector can be designed using this. Here, the partial response means to extract information from adjacent data symbols to recover one symbol. In other words, it means that one data symbol affects several adjacent data symbols while passing through the channel, and this characteristic can be utilized for data symbol restoration.

The characteristics of optical channels are determined by the density of data symbols, the spacing between tracks, the characteristics of the optical pickup, and the characteristics of the modulation scheme used. In fact, it is almost impossible to model a transfer function that completely includes these characteristics. Even if it is possible, it is almost impossible to create a circuit that completely restores a signal with this complex transfer function. Therefore, it is most desirable to find a transfer function that is most similar to the characteristics of the channel but to compensate the difference between the modeled characteristic and the actual characteristic to some extent through the equalizer. System (partial response maximum likelihood (PRML) system).

FIG. 1 is a block diagram of a PRML system to which the Viterbi detection apparatus of the present invention is applied. A block surrounded by a dotted line shows a Viterbi detection apparatus 100 according to the present invention. 1, the analog signal output from the optical pickup 2 is converted into a digital signal by the A / D converter 4, and the converted digital signal is converted into a digital signal by the characteristic of the actual channel and the target model characteristic of the Viterbi detection device 100 Is inputted to the Viterbi detector 100 through an equalizer 6 for compensating for the difference between the Viterbi detector and the Viterbi detector. In this case, as shown in Fig. 1, the output of the equalizer 6 may also be output as a signal not passed through the Viterbi detection device 100 and used for other purposes.

The Viterbi detection apparatus 100 according to the present invention includes a register 10, a branch metric calculator (BMC) 20, an add-compare-select unit (BMC) (ACS) 30, an overflow controller (OVC) 40, a state metric memory (SMM) 50, and a path memory (PM) The Viterbi detector 100 is substantially similar in structure to a general Viterbi decoder, however, the Viterbi detector of the present invention detects the original data stream from channel data that is not intentionally encoded based on channel modeling, It differs from non-decoder. Since the Viterbi detection device 100 for optical channels has a very simple transfer function, a small number of states, and a small hardware size, but requires a high clock rate and continuous output data, the register trace memory 60 And an overflow control logic 40 for preventing an overflow of the state evaluation amount.

The optical channel modeling applied to the present invention will be described below. Fiber channels are characterized by pit-land density, track density, pickup performance, and modulation transfer function. As described above, it is almost impossible to actually produce a complete model of the optical channel, and even if it is possible to fabricate a detector having a complete transmission function, it is hardly feasible. Therefore, the optical channel model having the transfer function that compensates for the difference between the actual model and the evaluation model using an equalizer that is closest to the channel characteristic and is not complicated can be modeled by a simple convolutional decoder as shown in Equation (1) .

F (z) = 1 + kz + kz 2 + z 3 ( where, k is a positive real number)

In this model, the value of k depends on the characteristics of the optical pickup and the state of the optical disc. The choice of the k value should be determined by the ratio of the 3T signal (T is the period of the signal) to the 4T signal of the signal input to the actual Viterbi decoder. Here, the 3T signal is a signal in which the data stored in the disk does not change over from 3T to +1, or from -1 to +1. The relation between this ratio and k should satisfy the following equation (2).

Level _3T : Level _4T = k: k + 1

Since the data to be written to the disk is only 1 and -1, the output of the channel having the transfer function of Equation (1) can have five signal levels (+ high, + middle, 0, -middle, -high) As a result, this value becomes the reference value in the BMC 20 in Fig. The middle value is set to the 3T signal level, and the ± high value is set to the 4T signal level.

In the present invention, transfer functions for various optical channel models are used, which are shown in Table 1, and their frequency response characteristics are shown in FIG.

Model Transfer function When k = 1.0, PR (1,1,1,1) F (z) = 1 + z + z 2 + z 3 When k = 2.0, PR (1,2,2,1) F (z) = 1 + 2z + 2z 2 + z 3 When k = 3.0, PR (1, 3, 3, 1) F (z) = 1 + 3z + 3z 2 + z 3

As shown in Fig. 8, there is no significant difference in the frequency domain, but there is a significant difference between the partial response characteristics in the time domain.

6A, 6B and 6C show the ratio of the 3T and 4T signals according to the transfer function of the optical channel model, and each point represents a time point at which quantization is performed and a signal size at that time. As can be seen from FIGS. 6A, 6B and 6C, the 3T level and the 4T level are different from each other, and the PR (1, 1) model is insufficient for explaining the characteristics of the channel, It does not.

The transfer functions above have a constraint length of 4, so the number of states is 8, which is 2 ^k-1 . However, data coded by the EFM method is limited to a signal switching interval of 3T or more. Therefore, the Trellis diagram of the optical channel based on this characteristic is as shown in FIG. Here, the reference value should be scaled to fit the range of the input signal. The shape of the trellis diagram of PR (1,1,1,1) and PR (1,3,3,1) is also similar. Due to the EFM characteristics, the number of states is two, The number of branches decreased by eight. As a result, the minimum distance between the symbols increases and the error correction capability increases.

Referring again to FIG. 1, the method and apparatus for detecting a Viterbi according to the present invention will be described in more detail.

The Viterbi detection method of the present invention first determines the optical channel model to be decoded according to the transfer function of the optical channel model. Reads out data read from the optical pickup, and calculates a difference absolute value between the data and each status reference value. The Euclidean distances are respectively obtained by adding the two previous state evaluation quantities and the branch quantities transitioned to the current state on the basis of the respective quantities calculated using the calculated difference absolute values, And a small value of the two values is selected to determine a new state evaluation amount and a path selection signal is generated. If the number of states that transition to one state is one, the result of adding the previous state estimation amount and branching amount is determined as a new state estimation amount. And converges the difference between the largest value and the smallest value of the determined state estimation amount to a predetermined value. And stores the determined overflow-protected new state estimate and sends the stored output to a previous state estimate of the add-compare-select step. Receives the path selection signal from the add-compare-select step, and restores and outputs the original signal using the register exchange structure. In this embodiment, the model to be decoded is changed according to the setting of the register value, and is configured to be able to detect Viterbi of transfer functions of various optical channel partial response models.

The BMC 20 of the Viterbi detection apparatus 100 of the present invention calculates each branch metric with input data. That is, the BMC 20 receives the data read from the optical pickup and calculates a difference absolute value between the data and each status reference value. The absolute value of these two values is the Euclidean distance. In this case, in the present invention, the optical channel has the transfer function of Equation (1) and is designed to calculate the branching amount with respect to the k value. An example of the BMC 20 includes five reference levels (+ high, + middle, and so on) of a reference value converter 22 and input data and a state reference value so as to enable decoding for a plurality of channel models, 0, -middle, -high) and outputting the absolute value calculator 242, 244, 246, 248, and 250. In this structure, the distribution has a width of 5 bits and ranges from +15 to -16. However, as a problem of the reliability range of the A / D converter 4, it is limited to be calculated in the range of +12 to -12. Here, the + high value is set to 12, the -high is set to -12, and the change of k is the change of the values of + middle and-middle, so that the gain of the input signal of the A / D converter 4 is adjusted as k Should be able to. This gain should be adjusted until the peak level of the 4T signal becomes a + high value. As shown in Table 2 below, four transfer functions are used in the present invention.

middle level k 6 1.0 7 1.4 8 2.0 9 3.0

The levels higher than the peak of the 4T signal never occur in Equation (1), but the peak value of the signals longer than 5T actually occurs to be larger than the 4T signal peak value. The levels higher than the peak of this 4T signal are meaningless in terms of the transfer function and must be clipped. Prior to the branching calculation, levels higher than 12 are clipped to 12, and levels lower than -12 are clipped to -12.

The register 10 sends the stored register value to the Viterbi detection device 100 to determine a model to decode. A general Viterbi detector performs a function only for one target model. However, as described above, the Viterbi detector of the present invention has a register for determining and storing one of a plurality of target models as a register value, (Register value? K), and the decoding target model is changed in accordance with the value k in Equation (1).

The ACS 30 typically operates in three steps. Based on the calculated amount of the input using the input data, the ACS 30 adds the two previous state estimates and branch amounts that transition to the current state, Respectively. Next, the two added values are compared with each other, and a smaller one of the two values is selected to determine a new state evaluation amount. The determined state estimate 34 is passed to the SMM 40 and the selected branch information, the path selection signal 32, is passed to the PM 60. The ACS 30 can not have a pipeline structure because the previous result must be reflected in the current calculation right away. Therefore, it should operate within only one clock period. The operation speed becomes slower as the bit width of the state estimation amount becomes longer. The minimum bit width of the state estimation amount is appropriately determined by the bit width of the branch amount and the overflow control logic to be described later. As shown in FIG. 2, the ACS 30 is used only in the states (-1, -1, -1) and (+ 1, + 1, + 1) by the RLL characteristic of the EFM code.

The SMM 50 stores the new state estimate generated by the ACS 30. The status evaluation memory consists of flip-flops.

OVC 40 prevents overflow of the state estimate memory. The Viterbi detection device must process continuously for infinite input data. The state estimates continue to increase because the previous state estimates and the selected estimates continue to accumulate. The state evaluation value converges to a certain value without increasing the difference between the largest value and the smallest value indefinitely. This is because it always selects a smaller value from the two state estimates that transition to one state. The difference between the state estimation amount having the largest value and the state estimation amount having the smallest value among the operations of the Viterbi detector having the Trellis diagram of FIG. 2 satisfies the inequality (3) shown below.

SM _max -SM _min < 2 ^{a + 2} (where a is the bit width of the branch quantity)

The overflow control in the Viterbi detector requires high-speed operation since it must occur together with the ACS operation every clock. In the present invention, a bitwise AND gating method which can operate at a high speed as shown in FIG. 3 is used.

If the expression (3) is satisfied and the number of bits of the state estimation amount is set to a + 2, a time point at which the most significant bit of all state estimation amounts becomes 1 necessarily occurs before overflow occurs. Thus, in this case, setting it all to zero yields the same result as subtracting 2 ^{a + 2} for all state estimates. That is, when a = 5, a case where all the most significant bits (MSBs) before the overflow occurs necessarily occurs. In this case, as shown in FIG. 3, When all the input values are 1, 0 is output to prevent overflow. The remaining bits except the MSB are transmitted to the SMM 50 using the NAND gate signal and the respective AND gates. Here, the SMM 50 has six bits of SMM (1), SMM (2), ..., and SMM (6) in the case of 8 bits. If the OVC 40 is located between the input terminal of the ACS 30 and the output terminal of the SMM 50, the number of bits of the state quantity is increased by one, but the two problems of simplification of the hardware and high- have.

The PM 60 basically uses a register exchange structure, but the RLL characteristic of the EFM modulation code can remove some states and branches in a general branch structure. The structure of the PM is as shown in Fig. 4, and it is generally known that the length is close to the theoretical performance when it is used about 5-6 times of the constraint length k. Here, the PM length is set to 24 based on 6 times of k. However, since there is a state in which the ACS is removed due to the removal of some branches in practice, the data stored in the PMs in the previous three stages are always constant, and this portion is removed. Also, at the final stage of the PM, unnecessary parts are removed since all the data of the states are converged into one value. 5A and 5B show respective cell structures constituting the PM. As shown in FIG. 4, a cell where two branches meet is shown in FIG. 5A, and a cell to which only one branch is connected is shown in FIG. Respectively. A multiplexer 62 for receiving the upper branch signal and the lower branch signal and selectively outputting the upper branch signal and the lower branch signal according to the path selection signal 32 and a multiplexer 62 for receiving the output of the multiplexer 62 and receiving the output of the multiplexer 62 under the control of the clock CLK And a flip-flop 64 for outputting stored data. The cell structure having only one branch has a flip-flop 66 for storing and outputting a branch signal in accordance with a control clock without a multiplexer.

Although it has been described above that the position of the OVC 40 in the Viterbi detector structure is between the output of the ACS 30 and the input of the SMM 50, And it can be designed to be located between the input terminal and the output terminal of the SMM 50. The operation description in this case is similar to the other description and therefore will be omitted. However, in this case, the bit width of the SMM 50 must be increased by one. Also, in OVC 40, the AND gate may take other bits as well as the most significant bit.

As described above, the Viterbi detection method and apparatus according to the present invention can be applied to a high-speed optical storage medium, and can detect an EFM-modulated signal using a simple equalizer. Due to the RLL characteristic of EFM modulation, Since the branches can be removed, there is an effect of reducing hardware. The use of the Viterbi detector of the present invention makes it unnecessary to use a sophisticated equalizer, thereby reducing the manufacturing cost and increasing the operation speed by using the overflow controller. In addition, since Viterbi detection can be performed for various channel models, it is possible to efficiently cope with channel changes.

Claims (16)

A Viterbi detection method applied to a signal restoration apparatus for a high-speed optical disc, (a) determining a channel model to be decoded; (b) receiving the read data from the optical pickup and calculating a difference absolute value between the data and each status reference value; (c) calculating the Euclidean distance by adding the two previous state evaluating quantities and the branch quantities which are transited to the current state, respectively, on the basis of the respective quantities calculated using the calculated difference absolute values, and Selecting a smaller value among the two values to determine a new state evaluation amount and generating a path selection signal; (d) preventing an overflow using the fact that the difference between the largest value and the smallest value of the determined state estimation amount converges to a predetermined value; (e) storing the determined overflow-protected new state estimate and sending the stored output to the previous state estimate in step (c); And (f) receiving a path selection signal from the step (c), and removing a predetermined number of states and branches using a register exchange structure. The method of claim 1, wherein in step (a), a decoding target model is determined and stored according to transfer functions of various optical channel partial response models. 2. The method of claim 1, wherein, in step (d), a is converged to 2 ^{a + 2} when a is a bit width of a branch quantity. The method of claim 1, wherein the step (d) comprises bitwise and gating the most significant bit or other bits. A Viterbi detection method applied to a signal restoration apparatus for a high-speed optical disc, (a) determining a channel model to be decoded; (b) receiving the read data from the optical pickup and calculating a difference absolute value between the data and each status reference value; (c) calculating the Euclidean distance by adding the two previous state evaluating quantities and the branch quantities which are transited to the current state, respectively, on the basis of the respective quantities calculated using the calculated difference absolute values, and Selecting a smaller value among the two values to determine a new state evaluation amount and generating a path selection signal; (d) storing the determined overflow-protected new state estimate and transmitting the stored output; (e) preventing overflow by using a difference between a largest value and a smallest value of the stored state estimation amount converged to a predetermined value, and outputting the result to the previous state evaluation amount of the step (c); And (f) receiving a path selection signal from the step (c), and removing a predetermined number of states and branches using a register exchange structure. 6. The method of claim 5, wherein the step (a) determines and stores a decoding target model according to transfer functions of various optical channel partial response models. 6. The method of claim 5, wherein in step (e), a is converged to 2 ^{a + 2} when a is a bit width of the quantization step. 6. The method of claim 5, wherein the step (e) comprises bitwise and gating the most significant bit or other bits. A Viterbi detecting apparatus applied to a signal recovery apparatus for a high-speed optical disc, A register for determining and storing the optical channel model to be decoded; A division calculation unit for receiving the read data from the optical pickup and calculating a difference absolute value between the read data and each state reference value; The Euclidean distances are respectively obtained by adding the two previous state evaluation quantities and the branch quantities transitioned to the current state on the basis of the respective quantities calculated using the calculated difference absolute values, An add-compare-select unit for comparing the two values and selecting a smaller value among the two values to determine a new state evaluation amount and generating a path selection signal; An overflow control unit for preventing an overflow by using a difference between a largest value and a smallest value of the determined state estimation amount converged to a predetermined value; A state evaluation memory storing the determined overflow-protected new state estimate and transmitting the stored output to a previous state estimate of the add-compare-select; And And a path memory for receiving a path selection signal from said add-compare-selection unit and removing a predetermined number of states and branches using a register exchange structure. 10. The Viterbi detector of claim 9, wherein the register unit determines and stores a decoding target model according to transfer functions of various optical channel partial response models. 10. The Viterbi detector of claim 9, wherein the overflow controller converges to 2 ^{&lt; a + 2 &gt;} when a is a bit width of the branch quantity. 10. The Viterbi detector of claim 9, wherein the overflow control unit bitwise and gates the most significant bit or other bits. A Viterbi detecting apparatus applied to a signal recovery apparatus for a high-speed optical disc, A register for determining and storing the optical channel model to be decoded; A division calculation unit for receiving the read data from the optical pickup and calculating a difference absolute value between the read data and each state reference value; The Euclidean distances are respectively obtained by adding the two previous state evaluation quantities and the branch quantities transitioned to the current state on the basis of the respective quantities calculated using the calculated difference absolute values, An add-compare-select unit for comparing the two values and selecting a smaller value among the two values to determine a new state evaluation amount and generating a path selection signal; A state evaluation memory part for storing the new state estimation amount and transmitting the stored output; An overflow controller for preventing an overflow by using the fact that the difference between the largest value and the smallest value of the transmitted state estimate converges to a predetermined value, and then transmitting the over state to the previous state evaluation amount of the add-compare-selection unit; And And a path memory for receiving a path selection signal from said add-compare-selection unit and removing a predetermined number of states and branches using a register exchange structure. 14. The Viterbi detector of claim 13, wherein the register unit determines and stores a decoding target model according to transfer functions of various optical channel partial response models. 14. The Viterbi detector of claim 13, wherein the overflow controller converges to 2 ^{&lt; a + 2 &gt;} when a is a bit width of the branch quantity. 14. The Viterbi detector of claim 13, wherein the overflow control unit bitwise and gates the most significant bit or other bits.