JP2000165260A - Decoder - Google Patents

Abstract

PROBLEM TO BE SOLVED: To enhance efficiency and reliability of information decoding. SOLUTION: A decoder of a soft decision input soft decision output configuration is composed to a channel decoder 22 and an information resource decoder 23. When a fault detector 24 detects a fault in the process of information source decoding, an input signal stream for information source decoding is corrected on the basis of a likelihood signal denoting the reliability of the input signal stream for information source decoding, that is. That of an output signal stream for channel decoding, and decoding of the information source is repeated. Furthermore, a likelihood signal of an output of the channel decoder is corrected on the basis of fault detection in the process of information source decoding and the corrected likelihood signal is used for a likelihood signal for an input to the channel decoder so as to repeat channel decoding and also information source decoding.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a decoding device used for a digital information communication system or the like.

[0002]

2. Description of the Related Art In digital information communication, information source coding and decoding, channel coding and decoding, and modulation and demodulation are main elemental technologies. FIG. 5 shows a configuration of a general digital information communication system. On the sending side,
In the information source coder 1, "0", "1"
Is assigned. There are various types of information to be transmitted, such as audio, video, text, etc. Here, "0", "1", "2", "
An information sequence of 8 samples consisting of 4 types of integers of 3 ":
The case where 01001032 is transmitted will be described as an example.

[0003] In the information source encoder 1, variable code length is assigned to assign a short code word to information with a high appearance frequency and assign a long code word to information with a low appearance frequency. Here, as an example, "0" is "0", "1" is "10", "
It is assumed that a codeword of “110” is assigned to “2” and a codeword of “1110” is assigned to “3.” Therefore, from the information source encoder 1, an information source encoded bit string; 0: 10: 0: 0: 0: 10: 0 ： 1110 ： 11
0 is output. At this time, the information source coded bit string becomes data of 15 bits in total. In the above description, “:” indicating the timing is inserted between the codewords for each information for the sake of simplicity for the sake of simplicity. In order to decode the above-mentioned information source coded bit string on the receiving side, it is necessary to know the head position of the coded bit string. Therefore, a synchronization pattern "11110" is added before the information source coded bit string, and , The information source coded bit string with the synchronized pattern is: sync pattern + information source coded bit string; 11110: 0: 10:
This becomes 20-bit data of 0: 0: 10: 0: 01110: 110.

[0004] Next, in order to detect or correct an error on a receiving bit due to a noise, an interference wave, fading, or the like in a transmission path, the 20-bit signal is used. The data sequence is input to the channel encoder 2, and a parity bit is added based on an appropriate encoding rule. Here, as an example, assuming that a systematic convolutional code having a coding rate of 1/2 is used, a 20-bit parity bit is added to the 20-bit information source coding bit sequence, and a communication channel coding bit sequence of 40 bits in total. Is generated.

Next, in order to convert a transmission bit into a signal that meets transmission conditions while considering transmission efficiency and reliability,
Input to modulator 3. Here, as an example, two-phase Phase Shift Keying (2-PSK) is used as a modulation method, and "1" is used.
Are assigned to a signal value of "+1" and a bit of "0" is assigned to a signal value of "-1". At this time, the 2-PSK signal for the synchronization pattern + information source coding bit string is a 2-PSK signal; +1, +1, +1, +1, -1, -1, +1, -1, -1, -1 -1,
+1 -1, -1, -1, +1, +1, +1, -1, +1, +1, -1, followed by a 2-PSK signal for the parity bit. The output signal of the modulator 3 is transmitted to a transmission path. On the receiving side, a signal to which noise is added via a transmission path is input, demodulated by a demodulator 4, error-corrected by a channel decoder 5 based on the parity bits, and then decoded by an information source decoder 6. To reproduce information.

By the way, as a method of the above-mentioned channel decoding,
At the previous stage of the decoder, the signal is set to "1" depending on whether or not the real-valued signal output from the demodulator is greater than 0 level.
Or hard-decision input decoding that decodes after determining "0",
Also, soft-decision input decoding for decoding using a real-valued signal as it is is known. This soft-decision input decoding method is known to have better error correction characteristics than the hard-decision input decoding. Furthermore, there are various types of soft-decision input decoding. Among them, an input likelihood signal indicating the reliability of the input signal to be decoded is added, and the soft-decision input signal is subjected to maximum likelihood decoding. There is a method of outputting a decoded value together with an output likelihood signal indicating the reliability of the decoded value. In this system, it is possible to perform hard-decision output decoding that outputs the decoded output after determining it as "1" or "0", and soft-decision output decoding that outputs the decoded value as a real value. The method using the decoder having the soft-decision input / soft-decision output configuration is described in detail in, for example, the following document. J. Hagenauer, E. Offer, and L. Papke, "Iterative deco
ding of binary blockand convolutional codes ", IEEE
Trans. Information Theory, vol.42, No.2, pp.429-44
5, March 1996.

Here, the meaning of the likelihood in the above-mentioned document will be described with reference to an example. For example, it is assumed that one signal value in the received signal sequence is “+0.6”. this is,
(1) When the transmission signal value is "+1.0" and the noise value "-0.4" is added and the reception signal value becomes "+0.6", (2) When the transmission signal value is "-1.0", When the noise value “+1.6” is added and the received signal value becomes “+0.6”, there are two possible cases, but it is not possible to determine which one is from the received signal value alone.
If a hard decision is made based on whether the received signal value is greater than 0, the decision value is "1". In the case of the above (1), the judgment result is correct, but conversely, in the above (2)
In the case of, the judgment result is wrong.

When channel decoding is performed based on the maximum likelihood decoding shown in the above document, a likelihood signal indicating the reliability of the decoded value is output. Here, the likelihood signal is defined by a likelihood = log (probability that the transmission signal value is “+1” / probability that the transmission signal value is “−1”) with respect to the reception signal value. That is, the higher the probability that the transmission signal value is “+1”, the larger the likelihood signal becomes, the larger the positive value.
The higher the probability that the transmission signal value is "-1", the larger the likelihood signal becomes. This soft-decision input, soft-decision output maximum likelihood decoder inputs an input likelihood representing the reliability of each signal constituting the received input signal block together with the real-valued received input signal block, An output likelihood representing the reliability of the signal is calculated and output, and the decoded output is: decoded output = received input signal + input likelihood + output likelihood. By using such a decoder, excellent error correction characteristics can be obtained.

[0009]

In the digital information communication system described above, whether or not there is a bit error in the decoded input at the time of decoding the information source has a great influence on the decoding quality. For example, when the above information is transmitted, the information source decoded input bit sequence excluding the synchronization pattern is the information source decoded input bit sequence: 0 1 0 0 0 1 0 0 1 1 1 0 1
When there is no error, such as 10, it is possible to generate a decoded information sequence;

However, the seventh bit erroneously changes from “0” to “1”, and the information source decoding input bit string; 0 1 0 0 0 1 1 0 1 1 1 0 1
In the case of 10, when a delimiter is inserted according to the above four types of codewords, an information source decoding input bit string; 0: 10: 0: 0: 110: 1110: 11
0, and the reproduction information is a decoded information sequence; That is, up to “0 1 0 0” up to the fourth sample is correct, but not after the fifth sample, and seven samples are generated in total. In this case, the seventh bit of the received bit string was erroneous from “0” to “1”, but at that time it was impossible to detect whether or not the error was incorrect. Only when it is input to the decoder and decoded to detect that there are only 7 samples of decoded information, it is detected that there is an error in the received bit and that the decoding of the information source is abnormal. Then, it cannot be determined how far the decoded information sequence is reliable.

Also, the 8th bit is erroneously changed from “0” to “1”, and the information source decoding input bit sequence: 0 1 0 0 0 1 0 1 1 1 1 0 1
In the case of 10, according to the four types of codewords, if a delimiter is inserted, the input source decoding input bit string will be 0: 10: 0: 0: 0: 10: 11111110, and there is a pattern “1111” that is not in the codeword Decoding is stopped, and the reproduction information is a decoded information sequence; 0 1 0 1, which is correct up to 5 samples up to “0 1 0 0 1”, but 6
After the sample is not reproduced, five samples are generated in total. In this case, the 8th bit of the received bit string was erroneous from “0” to “1”, but at that time, it was impossible to detect whether the error was erroneous, and the 11th bit was not included in the code word. ” The pattern "1111" is detected, and it is detected that there is an error in the received bit and that the decoding of the information source is abnormal. Then, it cannot be determined how far the decoded information sequence is reliable.

In variable-length coding, decoding is performed while reproducing the timing. Therefore, if there is an error in the decoded input bit, the timing is erroneous, and the code error may spread to subsequent decoding or become impossible to decode. There are many. Deterioration of decoding quality due to such a code error generally indicates that the higher the compression ratio is,
Become noticeable. As described above, as a result of the information source decoding, a predetermined number of samples are not reproduced, or a pattern that is not included in the code word is input. It was not possible to determine whether there was an error. Therefore, there is a problem that it is impossible to determine how reliable the decoded information sequence is. The above document describes only a channel decoding method based on likelihood signals, and does not describe a system combining channel decoding and information source decoding.

SUMMARY OF THE INVENTION It is an object of the present invention to provide a decoding apparatus which can determine the reliability of a decoded information sequence when an abnormality is detected in decoding an information source. Another object of the present invention is to provide a decoding device capable of performing more efficient and highly reliable information reproduction by strengthening the mutual relation between the decoding of the communication path and the decoding of the information source.

[0014]

In order to achieve the above object, a decoding apparatus according to the present invention is a decoding apparatus for decoding information which has been subjected to channel coding after information source coding, and comprising: Decoding means and information source decoding means, the communication path decoding means having means for outputting a likelihood signal of a signal constituting a decoding output block of the communication path decoding means, and the information source decoding means, Means for outputting an abnormality detection signal of a signal constituting a decoding output block of the information source decoding means,
The information source decoding means may include one or both of a likelihood signal of a signal forming the decoded output block of the communication path decoding means and an abnormality detection signal of a signal forming the decoded output block of the information source decoding means. A means for determining a likelihood of a signal constituting a decoded output block of the information source decoding means based on the information. Further, the communication path decoding means includes means for correcting a likelihood signal of a signal constituting a decoded output block of the communication path decoding means. Further, the channel decoding means has means for inputting a likelihood signal of a signal constituting a decoding input block of the channel decoding means, and correcting a signal constituting a decoding output block of the channel decoding means. The likelihood signal is used as a likelihood signal of a signal constituting a decoded input block of the channel decoding means.

Further, the decoding process in the communication channel decoding unit and the decoding process in the information source decoding unit are repeated. Further, the means for correcting the likelihood signal of the signal constituting the decoded output block of the channel decoding means comprises: converting the likelihood signal of the signal constituting the decoded output block of the channel decoding means to the information source decoding means. Is replaced with a likelihood value determined from the likelihood determination result of the signal constituting the decoded output block. Further, the means for correcting the likelihood signal of the signal constituting the decoded output block of the channel decoding means comprises: converting the likelihood signal of the signal constituting the decoded output block of the channel decoding means to the information source decoding means. Is replaced by a signal corresponding to a known synchronization signal pattern included in the decoded input block. Further, the means for determining the likelihood of a signal constituting a decoded output block of the information source decoding means detects a correlation between the decoded output block and a decoded output block in the vicinity thereof,
According to the detection result, the likelihood of a signal constituting a decoded output block of the information source decoding means is determined.

[0016]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS (First Embodiment of the Present Invention) A first embodiment of the present invention will be described with reference to FIG. FIG. 1A shows a configuration on the transmission side, where an information source coder 11, a channel coder 12, and a modulator 13 are provided.
Have the same configuration as the information source coder 1, the communication channel coder 2, and the modulator 3 in FIG. (B) shows the configuration of the receiving side, and the demodulator 2
1 has the same configuration as the demodulator 4 in FIG. 2
Reference numeral 2 denotes a channel decoder, which is a decoder having a soft-decision-input soft-decision output configuration configured to output the output likelihood signal together with the soft-decision output signal. Reference numeral 23 denotes an information source decoder for decoding the channel decoding output (that is, the information source decoding input) output from the channel decoder 22 as an information source. This information source decoder 23 also has a soft decision input soft decision output configuration. And the decoder. Reference numeral 24 denotes an abnormality detector for detecting an abnormality in the decoded information output from the information source decoder 23. Reference numeral 25 denotes an abnormality output from the information source decoder 23 based on the outputs of the abnormality detector 24 and the channel decoder 22. It is a determining means for determining the likelihood (reliability) of the decoded output signal.

Regarding the operation of the embodiment of the present invention thus configured, "0", "0", as in the case described above.
An example in which an information sequence information sequence of eight samples consisting of four types of integers of 1 "," 2 ", and" 3 "; 01001032 will be described as described above. Encoding is applied and a synchronization pattern is added,
Synchronization pattern + source coding bit string; 11110: 0: 10:
20-bit data of 0: 0: 10: 0: 1110: 110 is input to the channel encoder 12. Assuming that a systematic convolutional code having a coding rate of 1/2 is used as a channel coding method, a 20-bit parity bit is added to the 20-bit data sequence, and a total of 40-bit channel coding bit sequences are obtained. Generated. Next, in the modulator 13, two-phase phase shift keying (2-PSK) is used as an example.
The bit of 1 is assigned to the signal value of “1” and the bit of “0” is assigned to the signal value of “−1.” At this time, the 2-PSK signal for the synchronization pattern + information source coding bit sequence is a 2-PSK signal; , +1, +1, +1, -1, -1, +1, -1, -1, -1,
+1 -1, -1, -1, +1, +1, +1, -1, +1, +1, -1, followed by a 2-PSK signal for the parity bit.

Signals received via a communication path are generally real-valued signals containing noise. For example, the received signal of the 2-PSK signal corresponding to the above-mentioned synchronization pattern + information source coded bit string is a received 2-PSK signal; +0.8, +1.2, +0.6, +1.0, -0.4, -0.8,
+0.2, -0.8, -1.4, -1.0, +0.4, -1.2, -1.0, +0.6, +
Suppose they were 0.4, +0.4, -0.2, +0.6, +1.0, -0.4. This is followed by the received 2-PSK signal for the 20 parity bits. A 2-PSK demodulated signal corresponding to the 40-bit signal is output from the demodulator 21 and input to the channel decoder 22. Here, the channel decoder 22 is a decoder having a soft-decision input / soft-decision output configuration, as described above, and outputs a soft-decision decoded output corresponding to the signal of the 20-bit synchronization pattern + information source coded bit string. A signal is output with a corresponding output likelihood signal. The channel decoding output from the channel decoder 22 is a highly accurate determination result because the output likelihood is considered.

A method for improving the accuracy of the decision result in the channel decoder 22 will be described with reference to the following five cases of output likelihood. When one signal value in the above-described received signal sequence is “+0.6”, the following five cases of output likelihood are shown as examples. (1) Assuming that the output likelihood of the decoded value is output as “+0.8”, the decoded value of the soft decision output is “0.6 + 0.8 = 1.4”, and
The hard decision output is “1”. (2) If the output likelihood of the decoded value is output as “+0.2”, the decoded value of the soft decision output is “0.6 + 0.2 = 0.8”, and
In this case, the hard decision output is “1”. (3) If the output likelihood of the decoded value is output as "-0.4", the decoded value of the soft decision output is "0.6-0.4 = 0.2", and the hard decision output in this case is "1". (4) If the output likelihood of the decoded value is output as “−0.8”, the decoded value of the soft decision output is “0.6−0.8 = −0.2”, and
The hard decision output in this case is “0”. (5) Assuming that the output likelihood of the decoded value is output as "-1.4", the decoded value of the soft decision output is "0.6-1.4 = -0.8", and
The hard decision output in this case is “0”.

The above five examples will be compared. (1) Example 1 and Example 2 are cases where the received signal value and the sign of the output likelihood are equal, and the received signal value and the sign of the decoded value are also the same. In this case, the sign of the determination result is the same between when the received signal value is hard-decided as it is and when the decoded value is hard-decided. That is, it is determined that there is no error in the determination of the received signal value as it is, and no error correction is performed. The larger the absolute value of the output likelihood information is, the larger the absolute value of the decoded value of the soft decision output is, that is, the higher the reliability of the decoded value is in Example 1. (2) In Example 3, when the received signal value and the sign of the output likelihood are different and the sign of the received signal value and the decoded value are the same, the hard decision of the received signal value as it is and the hard decision of the decoded value are performed. Sometimes, the sign of the determination result is the same. Also in this case, it is determined that there is no error in the determination of the received signal value as it is,
Error correction has not been performed. However, both the output likelihood and the decoded value are
Since the absolute value is small, it means that the reliability of the decoding result is low. (3) Example 4 is different from Example 5 in that the received signal value and the sign of the output likelihood are different, and the received signal value and the decoded value are different in sign.
The sign of the determination result is different between the hard decision of the received signal value as it is and the hard decision of the decoded value, and it is determined that there is an error in the decision of the received signal value as it is, and error correction is performed. ing. In Example 4, since the absolute value of both the output likelihood and the decoded value is small, the reliability of the decoded result is low, which means that erroneous correction may have been performed. Example 5 indicates that both the output likelihood and the decoded value have a large absolute value, the reliability of the decoded result is high, and there is a possibility that error correction has been performed correctly. Thus, in the present invention,
Since a decoder having a soft-decision input and soft-decision output configuration that outputs a likelihood signal together with a decoded output is used as the channel decoder 22, it is possible to output a highly accurate channel decoded output.

Next, the communication path decoder 2
The channel decoding output corresponding to the signal of the synchronization pattern + the information source encoded bit string (20 bits in this example) output from 2 is input to the information source decoder 23. This information source decoder 23 is also configured by a decoder having a soft-decision input / soft-decision output configuration like the channel decoder 22.
The output likelihood signal output from the channel decoder 22 is input as the input likelihood together with the soft-decision decoded signal output of the channel decoder 22.

In the information source decoder 23, first,
The synchronization pattern is detected. That is, the Euclidean distance between the synchronization pattern signal and the soft-decision channel decoded signal is calculated by sliding the channel decoded signal one signal at a time, and the synchronization point is determined from the minimum Euclidean distance calculation value. Next, a signal sequence from the signal following the last signal of the detected synchronization pattern to before the first signal of the next synchronization pattern (in the above-described example, a signal sequence of 15 bits).
Is subjected to information source decoding as follows. In the information source decoding process of this signal block, first, a hard-decision of the soft-decision output channel decoding signal from the channel decoder 22 is performed, and a variable-length code Is decrypted.

Here, when the information source decoding input bit string is free of errors such as: information source decoding input string; 0 1 0 0 0 1 0 0 1 1 1 0 1 1 0 01001032 can be generated. On the other hand, the 7th bit is incorrectly changed from “0” to “1”, and the information source decoding input bit string; 0 1 0 0 0 1 1 0 1 1 1 0 1
In the case of 10, according to four kinds of codewords, if a delimiter is inserted, it becomes an information source decoding input sequence; 0: 10: 0: 0: 0: 110: 1110: 110, and the reproduction information is a decoding information sequence;
7 samples will be generated. On the other hand, when the 8th bit is erroneously changed from “0” to “1” and is the source decoding input sequence; 0 1 0 0 0 1 0 1 1 1 1 0 1 1 0, it follows four types of codewords. , Insert a delimiter, then the source decoding input string; 0: 10: 0: 0: 0: 10: 1 1 1 1 1 1 1 1 0, and there is a pattern "1111" that is not in the codeword, so decoding stops, The reproduction information is a decoding information sequence;
This results in the generation of 5 samples of 1.

The abnormality detector 24 detects an abnormality in information source decoding due to such a bit error in the following procedure. (1) Since the number of samples to be reproduced is often known, an abnormality in information source decoding is detected based on whether the number of reproduced samples is equal to a predetermined number. (2) An abnormality in information source decoding is detected based on whether or not the information source decoding is stopped due to detection of a pattern not defined in the codeword. (3) Information such as images and sounds often has a high correlation between an information signal and information signals around the information signal. Therefore, a correlation between the decoded information and the surrounding decoded information is calculated, and an abnormality in the information source decoding is detected based on whether or not the correlation is high. If no abnormality in the information source decoding is detected by such a procedure, the information source decoding for this information source decoding input bit sequence is terminated.

On the other hand, when an abnormality in information source decoding is detected, the judging means 25 judges the reliability of the communication channel decoding result as follows. (1) When the received signal value, the code of the output likelihood and the code of the decoded value are all positive in the channel decoding, if the output likelihood is larger than a predetermined value (for example, +0.5), the channel decoding is performed. The reliability of the result is determined to be high. Alternatively, when the signs of the above three signals are all negative, the output likelihood is a predetermined value (eg, -0.5)
If it is smaller than this, it is determined that the reliability of the channel decoding result is high. (2) The output likelihood at the time of channel decoding is a predetermined value (for example,
If it is larger than (+0.8), it is determined that the reliability of the channel decoding result is high. Alternatively, if it is smaller than a predetermined value (for example, -0.8), it is determined that the reliability of the channel decoding result is high. (3) For each signal that does not apply to (1) and (2) above, it is determined that the reliability of the channel decoding result is low. Although the output signal sequence of the channel decoding becomes the input signal sequence of the information source decoding, the likelihood of the output of the information source decoding, that is, the reliability is determined to be high until a signal of low reliability appears for the first time. In this way, it is possible to determine how much of the output signal sequence of the information source decoding is reliable.

(Second Embodiment of the Present Invention) Further, based on the above-mentioned determination result, the hard-decision result of a signal with low reliability in the output signal sequence of the channel decoding is sequentially inverted, and If the decoding is re-executed and the abnormality of the information source decoding is no longer detected, the inversion is determined to be correct, and the decoding can be continued. FIG. 2 is a diagram illustrating a configuration example of the receiving side according to the second embodiment of the present invention as described above. As shown in this figure, in this embodiment,
Correction means 26 is provided between the channel decoder 22 and the information source decoder 23. This correction means 26
The output of the channel decoder 22 is corrected based on the determination result of the determination unit 25.

For example, if the 8th bit of the information source decoding input string is erroneous from "0" to "1" as described above, there is a pattern "1111" which is not in the code word. Then, the abnormality is detected and the information source decoding stops. This is because such anomaly occurred because the input signal of the eighth information source decoding, that is, the hard decision signal of the output signal of the channel decoding was incorrect from "0" to "1". here,
When the input signal value of the eighth channel decoding is “+0.8”,
If the output likelihood is, for example, negative “−0.7”, the output value of channel decoding is “+ 0.8−0.7 = + 0.1”, the hard decision value is “+1”, and “0” is changed to “1”. " This is the case where the sign of the input / output signal for channel decoding and the sign of the output likelihood are different, and the absolute value of the output likelihood is large. In such a case, there is a high possibility that the hard decision value of the output of the channel decoding is incorrect.

Therefore, in such a case, the judging means 25 sequentially starts from the eleventh input signal in which the abnormality is detected, the input signal of the preceding information source decoding, that is, the output of the channel decoding. The output likelihood of each signal is determined retroactively. Then, the signal of the input / output signal of the channel decoding and the code of the output likelihood are different, and a signal having a large absolute value of the output likelihood is searched for. Invert the hard decision value. This means that the absolute value of the output likelihood of channel decoding is corrected to be large. Then, the information source decoding process is executed again on the corrected channel code output.
Thereby, since the error of the eighth signal has been corrected, the abnormality in the information source decoding is resolved, and the information source decoding can be continued.

As described above, according to the present invention, the decoding is performed while strengthening the mutual relation between the decoding of the channel code and the decoding of the information source code. Can be determined, and the output quality of the information source decoding can be improved.

(Third Embodiment of the Present Invention) When an abnormality in information source decoding is detected, the output quality of information source decoding is repeated by repeating the channel decoding process and the information source decoding process. A third embodiment of the present invention for further improvement will be described. FIG. 3 is a block diagram showing the configuration of the third embodiment of the present invention. In this figure, the same components as those in FIG. 1 are denoted by the same reference numerals, and description thereof will be omitted. As shown in FIG. 3, in this embodiment, a likelihood corrector 27 is provided. In this embodiment, the channel decoder 22 uses an input likelihood signal indicating the reliability of an input signal to be decoded.
When the abnormality detector 24 detects an abnormality in information source decoding, the determination unit 25 and the likelihood corrector 27
Is used to correct the output likelihood at the time of channel decoding, and the corrected likelihood is used as an input likelihood signal as the channel decoder 22.
As the input likelihood. Then, the communication path decoding and the information source decoding are performed again.

That is, the output likelihood at the time of channel decoding is corrected as follows, and the corrected likelihood is input to the channel decoder 22 as the input likelihood, and the channel decoding and information decoding are performed. Perform source decoding iteratively. (1) An output likelihood signal is determined from the detected synchronization pattern. For example, when the synchronization pattern is “11110”, the output likelihood signal related to the synchronization pattern is “+ v, + v, +
v, + v, -v "(for example, v = + 1.0). (2) If the sign of the received signal value, the code of the output likelihood signal, and the sign of the decoded value in channel decoding are all positive, the likelihood If the degree signal is larger than a predetermined value (for example, +0.5), the output likelihood signal for each signal is set to “v”, and if smaller, the output likelihood signal is set to “0”. When the sign of all three signals is negative, if the output likelihood signal is smaller than a predetermined value (for example, -0.5), the output likelihood signal for each signal is set to "-v". The likelihood signal is set to “0.” (3) If the output likelihood signal at the time of channel decoding is larger than a predetermined value (for example, +0.8), the output likelihood signal for each signal is set to “v” ( For example, if v = + 1.0) or smaller than a predetermined value (for example, -0.8), the output likelihood signal for each signal is set to "-v". Ma For not each signal,
The output likelihood signal is set to "0". (5) The channel decoding is executed again using the output likelihood signals 1 to 4 as input likelihoods for channel decoding. (6) Perform information source decoding again.

Then, the abnormality of the information source decoding described above is detected, and if no abnormality is detected, the information source decoding is terminated there. If an abnormality in the information source decoding is detected, the above-described channel decoding and information source decoding are repeated. If an abnormality in information source decoding is detected even after the predetermined number of repetitions, it is determined that there is no effect due to the repetition processing, and the processing ends. According to this embodiment, the information quality of the information source decoding can be further improved.

(Fourth Embodiment of the Present Invention) In the above-described embodiment, based on the detection of an abnormality at the time of decoding the information source, a signal which is estimated to have low reliability of the channel decoding output is modified. Thus, the idea is to execute the channel decoding and the information source decoding again. Conversely, based on the abnormality detection at the time of decoding the information source, the concept of correcting the signal which is estimated to have high reliability of the channel decoding output, and performing the channel decoding and the information source decoding again. A fourth embodiment of the present invention, in which is introduced, will be described with reference to FIGS. This embodiment can be realized by the same configuration as the configuration shown in FIG.

For example, as shown in (a), the first communication channel decoding is executed in a state where there is no input likelihood, and when the output signal is input to the information source decoder 23, the 100th signal input Then, it is assumed that an abnormality is detected in the information source decoding (b).
At this time, as described in the first embodiment, the signal of the 100th or less output of the channel decoding output is often erroneous, and it is not possible to specify which signal is erroneous.

However, it can be assumed that the channel decoded output signals before the 100th one have higher reliability. For example, it can be said that the first to tenth channel decoded output signals have higher reliability than the 21st and subsequent signals.
For example, if the signal values of the first to tenth signals are "+0.6, -0.5, -0.9, +0.2, +0.8, +1.0, -1.5, +0.5,-
0.7, +0.3 ", the judgment value is" +1, -1, -1, +1, +1, +1, -1, +1, -1, +1 " At this time, the likelihood corrector 27 sets the input likelihood of the channel decoder 22 to "+ v, -v, -v, + v, + v, + v,-. v, + v, -v, + v "and perform the second channel decoding (c). Here, v is a real value, for example, set to 1.0.
It is assumed that the reliability is high. Due to the above input likelihood of the channel decoder, if the second decoding result is improved and the error is more corrected than the first channel decoding result, the second communication When the path decoding signal is input to the information source decoder and the second decoding is performed, the abnormality may be eliminated, and the decoding characteristics are improved (d).

If the result of the second decoding is not improved as compared with the result of the first decoding of the channel, the input likelihood at the time of the second decoding of the channel is not sufficient. is there. At this time, as shown in FIG.
The input likelihood of channel decoding is generated by the same operation as described above with respect to 11 to 20 channel decoding outputs in addition to 10 to 10, and the third channel decoding is executed. The third decoding result is improved over the second channel decoding result,
If the error is more corrected, the third decoding of the channel is input to the information source decoder and the third decoding is performed. (F).

If the result of the third decoding is not improved compared with the result of the second decoding of the channel, the input likelihood at the time of the third decoding of the channel is not sufficient. is there. In this case, the same processing is further performed on up to 21 to 30 communication path decoded outputs.

Conversely, if the result of the third decoding is degraded compared to the result of the second decoding of the channel, as shown in FIG. It can be assumed that there is an error. At this time, the input likelihood at the time of the third communication channel decoding was not appropriate. In this case, the setting of the input likelihood was narrowed, and for example, 1 to 15 input The input likelihood is set for the channel decoding output up to and the third channel decoding is executed again (i). Alternatively, similarly to the first embodiment, it is possible to specify a signal for which a determination error is made from the relationship between the channel decoding input signal and the channel decoding output likelihood.

As described above, according to the fourth embodiment of the present invention, the input likelihood is determined from a signal estimated to have a higher reliability of the channel decoding output, and the channel decoding and the information source decoding are performed again. The idea is to implement and gradually improve the characteristics. In this method, the synchronization pattern at the time of information source decoding included in the channel decoding output can also be regarded as a signal estimated to be highly reliable in the channel decoding output when correctly detected by the information source decoding. This can also be set as the input likelihood of channel decoding in the same manner as described above.

In the above description, the case where variable-length coding is used as information source coding has been described. However, the present invention is not limited to this, and the case of information source coding / decoding based on fixed-length coding is used. The same can be applied to.

[0041]

As described above, according to the decoding apparatus of the present invention, the reliability (likelihood) of the decoded information sequence is determined when it is detected that the decoding of the information source is abnormal. It becomes possible. Further, the mutual relation between the decoding of the communication channel and the decoding of the information source is strengthened, and more efficient and highly reliable information reproduction can be performed.

[Brief description of the drawings]

FIG. 1 is a diagram illustrating a configuration example of a first embodiment of the present invention.

FIG. 2 is a diagram illustrating a configuration example of a second embodiment of the present invention.

FIG. 3 is a diagram illustrating a configuration example of a third embodiment of the present invention.

FIG. 4 is a diagram for explaining a fourth embodiment of the present invention.

FIG. 5 is a diagram illustrating a conventional technique.

[Explanation of symbols]

 1, 11 Information source encoder 2, 12 Communication channel encoder 3, 13 Modulator 4, 21 Demodulator 5, 22 Communication channel decoder 6, 23 Information source decoder 24 Abnormality detector 25 Judgment means 26 Correction means 27 Likelihood Corrector

 ──────────────────────────────────────────────────続 き Continued on the front page F term (reference) 5J065 AB01 AC02 AD10 AE02 AF02 AG05 AH21 5K014 AA01 BA10 EA01

Claims (7)

[Claims] 1. A decoding device for decoding information on which channel coding has been performed after information source coding, comprising: a channel decoding unit and an information source decoding unit. Means for outputting a likelihood signal of a signal constituting a decoded output block of the communication path decoding means, wherein the information source decoding means outputs an abnormality detection signal of a signal constituting the decoded output block of the information source decoding means The information source decoding means comprises: a likelihood signal of a signal constituting the decoded output block of the communication path decoding means; and an abnormality detection of a signal constituting the decoded output block of the information source decoding means. A decoding device, comprising: means for determining the likelihood of a signal constituting a decoded output block of the information source decoding means based on one or both of the signals. 2. The decoding device according to claim 1, further comprising means for correcting a likelihood signal of a signal forming a decoded output block of said channel decoding means. 3. The channel decoding means has means for inputting a likelihood signal of a signal constituting a decoding input block of the channel decoding means, and a signal constituting a decoding output block of the channel decoding means. 3. The decoding apparatus according to claim 2, wherein the corrected likelihood signal is used as a likelihood signal of a signal constituting a decoded input block of the channel decoding unit. 4. The decoding apparatus according to claim 1, wherein a decoding process in said channel decoding unit and a decoding process in said information source decoding unit are repeated. 5. A means for modifying a likelihood signal of a signal constituting a decoded output block of said channel decoding means, said likelihood signal of a signal constituting a decoded output block of said channel decoding means being said information source. 4. The method as claimed in claim 2, wherein a likelihood value determined from a likelihood determination result of a signal constituting a decoded output block of the decoding means is replaced.
3. The decoding device according to claim 1. 6. A means for modifying a likelihood signal of a signal constituting a decoded output block of said channel decoding means, said likelihood signal of a signal constituting a decoded output block of said channel decoding means being said information source. 4. The decoding device according to claim 2, wherein the decoding device replaces a signal corresponding to a known synchronization signal pattern included in a decoding input block of the decoding device. 7. A means for judging the likelihood of a signal constituting a decoded output block of the information source decoding means detects a correlation between the decoded output block and decoded output blocks in the vicinity thereof, and according to the detection result. 2. The decoding device according to claim 1, wherein the likelihood of a signal constituting a decoded output block of the information source decoding means is determined.