KR100200582B1 - Acquirement method and device of quality estimation information for viterbi decoding data - Google Patents

Abstract

The present invention is a case of using a Viterbi decoder to correct the error data contained in the received data in the apparatus for transmitting and receiving digital data.

The present invention relates to a method for measuring the reliability of decoded data by evaluating the quality of decoded data, and to a suitable apparatus.

A difference calculator for calculating and outputting a difference between two state evaluation quantities inputted from the addition comparison selector, a memory device for storing information on the quality evaluation in the previous state and information on the quality evaluation in the new state; And a decision unit for generating quality evaluation information for a new state from the ground value setting register, the difference value calculator, and the storage device.

Therefore, according to the present invention as described above, it has the advantage that the reliability of the Viterbi decoded data can be more accurately evaluated by supplementing the limit of error detection possessed by the CRC, and is an essential addition in the Viterbi decoding process. By using the result of the operation of the comparator, the need for the ning surplus data required by the CRC is eliminated, so that the amount of transmitted / received data does not increase and the capacity of the hardware does not need to be increased.

Description

A method for acquiring quality evaluation information of Viterbi decoded data and a device suitable therefor.

1 is a diagram showing the configuration of a general Viterbi decoder.

2 is a block diagram of an entire system for encoding and decoding a general digital signal.

FIG. 3 is a diagram illustrating an NRZI system based on the characteristics of the precoder and the magnetic channel shown in FIG.

4A to 4D are state diagrams showing signal characteristics different from those of the NRZI system shown in FIG.

5 is a diagram showing an algorithm for explaining a method of acquiring quality evaluation information according to the present invention.

6 is a diagram showing a trellis for a convolutional code having a code ratio of 1/2 and a constraint length K = 3.

7 is a diagram showing the configuration of an addition comparison selection circuit of a Viterbi decoder according to the present invention.

8 is a trellis diagram for a method for obtaining quality evaluation information according to the present invention.

9 is a view showing an apparatus for obtaining quality evaluation information according to the present invention.

The present invention relates to a method for displaying a substance of Viterbi decoded data and a device suitable therefor, and more particularly, to a device for transmitting and receiving digital data, in which a Viterbi decoder is used to correct error data included in received data. The present invention relates to a method for measuring the quality of decoded data by evaluating the quality of decoded data, and to a suitable apparatus.

The present invention is used together with the adder-compare-delect of the Viterbi decoder to show the reliability of the entire data decoded using the state metric and the base value while the Viterbi decoding is performed. It is commonly used as a device for providing a value for betting. In particular, it is used as a device for providing additional information on the reliability of the decoded data in the Viterbi decoder used in the code division multiple access (CDMA) digital data transceiver.

In the prior art, a cyclic redundancy chcek (CRC), which is a cyclic code used for error detection, is generally used to evaluate reliability of decoded data output from a Viterbi decoder. This determines the reliability of the data decoded by the Viterbi decoder by the CRC. That is, a certain number of surplus data is sent to the information data to be sent at the time of data transmission, and the receiving side determines the reliability of the decoded data by checking the CRC for the data after Viterbi decoding.

However, in the conventional technique, when only the reliability CRC for Viterbi decoded data is evaluated, the redundant data for the CRC must be transmitted together, thereby increasing the amount of data to be transmitted and received.

Also, since the error detection capability by CRC is determined by the number of surplus data added, the reliability measurement of Viterbi decoded data using CRC is limited by the error detection capability of the CRC used. Therefore, measuring the reliability of the data decoded using only the CRC has an insufficient disadvantage.

The present invention was created to solve at least some of the above-mentioned problems. Since the amount of transmission / reception data is unnecessary, the amount of transmission / reception data does not increase, and the state evaluation amount generated during the addition comparison selection process, which is an essential element in the Viterbi decoding process, is required. It is an object of the present invention to provide a method for displaying quality of Viterbi decoded data and other suitable apparatuses which can display the quality of decoded data and evaluate the reliability without requiring surplus data.

In the Viterbi decoder including the addition comparison selector according to the present invention for achieving the above object, the method for obtaining quality evaluation information on the decoded data is GOOD only when the previous quality evaluation information is in the 0 state. Setting the initial value to the BAD for information other than zero; BM _U and BM _C are the upper and lower giro evaluation amounts inputted to the addition comparison machine or at the corresponding state, and SM _U and SM _C are the state evaluation amounts inputted to the addition comparison selector. Judging whether the evaluation information is GOOD | (BM _U + SM _U )-(BM _C + SM _C ) |

If the result of the determination step is correct, the current quality evaluation information is GOOD, otherwise the current quality evaluation information is BAD;

The previous quality assessment information is characterized by including the step of updating to the modern quality assessment information.

Apparatus for obtaining quality evaluation information on complex data in a Viterbi decoder including an add comparison selector according to the present invention

A base value register in which a base value is set;

A difference calculator for calculating and outputting a difference between two valuation amounts inputted from the addition comparison selector;

Memory that stores information about the quality assessment in the previous state and information about the quality assessment in the new state

And a decision unit for generating quality evaluation information about a new state from the base value setting register difference calculator and the storage device.

Hereinafter, the present invention will be described in more detail with reference to the accompanying drawings.

FIG. 1 is a diagram showing the configuration of a general Viterbi decoder, which is composed of a giro evaluation circuit 100, an addition / comparison / precedent circuit 102, a pass memory 104, and a state memory 106.

Giro evaluation circuit 100 MS is a circuit for evaluating a giro evaluation amount based on an assumed value according to a state diagram by R (K) which is a received case and a system configuration. The evaluated evaluation amount is compared with the status evaluation amount for the previous survival path stored in the state memory 106 and the paths added according to the trellis path in the addition / comparison selection circuit 102 and input in an arbitrary state. A small state evaluation amount is selected, and this selected state evaluation amount is input to the state memory 106 to update the state evaluation amount of the state memory. The path selection signal for the path selected by the addition / comparison / selection circuit 102 is input to the pass memory 104. In the pass memory 104, the value to be decoded by the path selection signal is trellis-type pass memory. Proceeding to 104, the original signal is decoded.

2 is a block diagram of an entire system for encoding and decoding a general digital signal.

First, a digital signal is recorded on a magnetic tape corresponding to a channel via a precoder 200 that converts a non-return to zero inverter (NRZI) signal. The signal reproduced through the magnetic channel 202 is subjected to amplitude correction of the signal through the linear equalizer 204, and then converted into a digital signal by the analog / digital signal converter 206 to the Viterbi decoder 208. Is entered.

In the video decoder 208, the original signal is decoded from the received signal.

3 illustrates an MRZI system in which FIG. 2 models the characteristics of the precoder 200 and the magnetic channel 202.

Referring to FIG. 3, D (300a, 300b) represents a memory device representing a delay of a signal, '+ block 302 represents an adder and a tag represents an exclusive logic gate 304.

The input digital signal A (k) is precoded through an XOR 304 with a value B (k-1) that delays the output B (k) of the precoder 200 by a symbol interval and outputs B (k). This output B (k) is outputted to the reproduction signal R (k) through a channel having a channel characteristic of (1-D).

4A to 4B are state diagrams showing the characteristics of signals according to the characteristics of the NRZI system of FIG.

The input / output relationship according to the state change between the digital input signal A (k) and the output signal R (k) reproduced through the channel is shown. When the input signal is '-1' and the current state is S0 (k), the output is '0' and the next state is S0 (k + 1), the input signal is '1' and the current state is S0 (k) When the output is '2' and the next state is S1 (K + 1), and when the input signal is 'I' and the current state is S1 (k), the output is '-2' and the next state is S0 (k + 1). ), The input signal is '-1', the current state is S1 (k), the output is '0', and the next state is Si (k + 1).

As shown in the state diagram in FIG. 3 and FIG. 4A, the value of the valuation of the next state is determined by the calculation of the received value of the current state and the assumed value according to the state diagram. In other words, the evaluation value of the next state is the path evaluation amount for the two paths by adding the evaluation value of the current state to the value obtained by the branch evaluation value between the value R (k) received in the current state and the assumed value of the path. In comparison, the most optimal path is evaluated and selected as a survival pass.

In other words, the amount of evaluation of the next state for

S0 (k + 1) = min {SO (k) + (R (k) -0) 2, SL (k) + (R (k)-(-2)) 2} (1)

And the evaluation amount of the next state with respect to 1 state is

S1 (k + 1) = min {SO (K) + (R (K) -2) 2, S1 (K) + (R (k) -2) 2} (2)

Becomes

Therefore, the evaluation amount of the next state selects the smaller value of the evaluation amount by comparing the evaluation amounts of the two passes with each other. In the above formulas (1) and (2), parts common to each other may be omitted to simplify the expression.

SO (k + 1) = min {SO (k), S1 (k) + R (k) +1} (3)

SO (k + 1) = min {SO (k), -R (k) + 1, S1 (k)} (4)

It becomes

4B is when S1 (k + 1) = S1 (k), SO (k + 1) = SO (k) in Eqs. (3) and (4), and the probability is S1 (k) -SO ( k) + R (k) +1 ≥ 0, Si (k) -SO (k) + p (k) -1 ≤ 0, i.e. the above equation is -1 ≤ S1 (k) -SO (k) + When R (k) ≤ 1, degree 3B becomes a survival pass. Similarly, FIG. 4C is a survival pass when S1 (k) -S0 (k) + R (k) ≧ 1. 4D is a survival path when S1 (k) -S0 (k) + R (k) ≤ -1.

Here, when the difference evaluation amount of the present state is Δ (k), Δ (k) = S1 (k) -SO (k). Then, the difference evaluation amount of the next state for the survival pass of FIG. 4C is in the region Δ (k) + R (k) ≤ 1, and the evaluation amount Δ (k + 1) = -R (k of the next state in FIG. 4C ) +1. In addition, the difference evaluation amount of the next state with respect to the survival pass of FIG. 4D is when it exists in the area of (DELTA) (k) + R (k) ≥ 1, and the difference evaluation amount Δ (K + 1) = -R (k) -1 of the next state Becomes Similarly, the next state for the survival pass in FIG. 4B is when the evaluation amount is -1 ≦ Δ (k) + R (k) ≦ 1 in the region and the difference evaluation value becomes Δ (k).

As shown in FIG. 4C, when the previous pass is summed at the node at -1, Δ (k + 1) = -R (k) +1 and the signal reproduced through the channel contains the noise component n (k). Signal R (k) = B (k)-B (k-1) + n (k), which is R (k) = [A (k) + B (k + 1)] mod2-B (k- 1) + n (k).

4C shows a state diagram when B (k-1) = -1, and therefore R (k) = A (k)-(-1) + n (k), that is, R (k) = A (k) + 1 + n (k). Therefore, Δ (k + 1) =-(A9k) + 1) + 1 + n (k), and Δ (k + 1) = -A (k) + n (k).

On the other hand, in FIG. 4C, since the value of A (k) has values of +1 and -1, the value of A (k) is -1 for a path given from -1 to -1 and thus Δ (k + 1) = -1 + n (k), and for a given path from -1 to +1, the value of A (k) is +1, so Δ (k + 1) = -1 + n (k ).

In summary, since Δ (k + 1) = -R (k) + 1 = -1 + n (k), it means that R (k) = 2-n (k), that is, R (k) = 2. Since Δ (k + 1) = -R (k) + 1 = 1 + N (K), it means that R (k) = 0 -n (k), that is, R (k) = 0.

However, in Fig. 4C, Δ (k + 1) = -R (k) + 1 to +1, which constitute the difference evaluation amount, is a threshold level value for the assumed signal value and the threshold level of the reproduction signal. The value is perturbed in the magnitude of the amplitude due to the distortion of the channel characteristics, and it can improve the decoding performance of the signal by adaptively detecting the threshold level value and comparing it with the received signal to calculate the difference evaluation amount. Make sure

5 is a diagram showing an algorithm for explaining a method of acquiring quality evaluation information according to the present invention.

Set initial value (500 steps).

Here, if the previous quality evaluation information is in the 0 state, it is set to GOOD, and for the states other than the previous quality evaluation information is 0, the initial value is set to BAD.

From the initial value set in step 500, it is determined whether the quality evaluation information is GOOD and the base value of (BM _U + SM _{U _-} (BM _C + SM _C ) | is the base value.

Here, BM _U + BM _c is the difference and lower branch evaluation amount inputted to the add comparison selector in the corresponding state, and SM _U and SM _C are state evaluation amounts input to the corresponding add comparison selector.

If the judgment of step 502 is correct, the current quality evaluation information becomes GOOD (step 504), otherwise the current quality evaluation information becomes BAD (step 506).

Therefore, the previous quality evaluation information is replaced with the current quality evaluation information. (Step 508)

This process is repeated until the comparative addition operation is completed for the unknown end of Viterbi decoding and the final quality evaluation information is output as the final quality evaluation information about the growth path at the last stage.

6 is a diagram illustrating a trellis diagram of an encoder. The trellis diagram of a convolutional encoder having a code rate of 1/2 and a limit length K of 3 is illustrated. The decoding method of the convolution code generally uses a Viterbi algorithm.

FIG. 7 is a diagram showing the construction of a housekeeping comparison circuit of a Viterbi decoder according to the present invention. The first adder 700 for inputting an upper branch evaluating amount BMu and a state estimating amount SMu and a lower branch evaluating amount BMc And a second adder 702 for inputting the state evaluation amount SMu, and a comparator 704 and a first adder 700 and a second adder 702 comparing the results of the first adder 700 and the second adder 702. A selector 706 for determining a new state evaluation amount from the output value

Since the Viterbi decoding method aims to search the maximum likelihood sequence estimate path in Trellis, the path having the minimum distance compared to the reception sequence through the discrete memoryless channel, The Survivor path with the highest evaluation amount is searched.

An example Viterbi decoder having a trellis as shown in FIG. 6 will be described below.

In each stage, new state evaluation and survival condensation information are found by calculation of the addition comparison selection circuit shown in FIG. 7 from the giro evaluation amount and the state evaluation amount input to each state.

4 is a trellis diagram for a method for obtaining quality assessment information according to the present invention.

If the survival path of the thick line shown in FIG. 8 is selected, only the quality evaluation information for the state SO is FOOD at step (t + 3), and the quality evaluation information for the remaining states S1, S2, and S3 is It becomes BAD.

9 is a diagram showing an apparatus for obtaining quality evaluation information according to the present invention, the configuration of which is a base value setting register 900 for setting and outputting a threshold value, and an addition comparison selection circuit shown in FIG. A difference calculator 902 that calculates and outputs a difference between two state evaluation amounts A and B output from the memory, and stores the information on the quality evaluation in the previous state and the information on the quality evaluation in the new state. An element 906, a base value setting register 900, a difference calculator 902 and a decision circuit 904 for generating quality evaluation information for a new state from the memory element 906.

The operation of the apparatus for obtaining quality evaluation information according to the present invention shown in FIG. 9 configured as described above is as follows.

The addition thrust value in the addition comparison selection circuit (shown in FIG. 7), which is an essential requirement for Viterbi decoding, is inputted A and B into the difference calculator 902 of the circuit for obtaining quality evaluation information. The difference between and B is calculated and then transferred to the decision circuit 904. The decision circuit 904 outputs and stores the base value and the difference value calculator 920 set in the base value register 900. The quality evaluation information of the new state is calculated according to a comfort algorithm to obtain the quality evaluation information from the previous state quality evaluation information stored in step 906. The Trellis also repeats the states of the last stage. In the final step, the calculation result for the condition with the minimum state quantity evaluation is the quality evaluation information about the survival path to be obtained, and this is the information for evaluating the degree of lightning for the Viterbi-decoded data.

As described above, according to the present invention, it is possible to improve the reliability of the Viterbi decoded data by supplementing the error detection of the CRC, and it is an addition comparison that is essentially used during the Viterbi decoding process. By using the result of the operation of the selector, the redundant data required by the CRC is not required, and thus the amount of data transmitted / received does not increase, and thus it is not necessary to increase the capacity of hardware.

Claims (2)

In a Viterbi decoder including an add comparison selector, a method for obtaining quality evaluation information for decoded data includes GOOD only if the previous quality evaluation information is in a zero state, and the states other than the previous quality evaluation information are BAD. Setting an initial value to; When BM _U and BM _U are upper and lower giro evaluation amounts inputted to the addition comparison selector in the corresponding state, and SM _U and SM _C are state evaluation amounts inputted to the corresponding addition comparison selector, the previous quality evaluation from the set initial value Determining whether the information is GOOD and | (BM _U + SM _U ) _{_-} (BM _C + SM _C ) | base value; If the determination result is correct, the current quality evaluation information is GOOD, otherwise the current quality evaluation information is BAD; A method of obtaining quality assessment information comprising the step of updating previous quality assessment information with current quality assessment information. An apparatus for obtaining quality evaluation information on decoded data in a Viterbi decoder including an add comparator, comprising: a base value register having a base value set; A difference calculator for calculating and outputting a difference between two state evaluation quantities inputted from the addition comparison selector; A storage element for storing information on the quality evaluation in the previous state and information on the quality evaluation in the new state; And a decision-making unit for generating quality evaluation information on a new state from the base value setting register, the difference value calculator, and the storage device.