JP2000049625A - Error correction system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To improve a transmission error correction efficiency and to realize reliable reception. SOLUTION: In a viterbi decoding circuit 14, demodulation signals (I and Q) are inputted to first and second branch metric generation circuits 141 and 142, a soft judgment branch metric is generated from the demodulation signals in the first branch metric generation circuit 141 and a hard judgement branch metric for a special processing before and after termination time is generated in the second branch metric generation circuit 142. The outputs are inputted to a selector 143 and the hard judgment branch metric is selected at the time of a termination processing and the soft judgment branch metric at the time of a regular processing at timing when a synchronizing signal protected from a synchronous byte detection result at the head of a transport stream packet is set to be a reference. In an ACS(addition/comparison/selection) circuit 144, a state metric following the trellis of codes is calculated by using the selected branch metric, and a path memory circuit 145 is controlled by the calculation result. Thus, a maximum likelihood path is traced.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] 1. Field of the Invention [0002] The present invention relates to a Viterbi decoding method for decoding a convolutional code that has been subjected to termination processing, and an error correction method using a Viterbi decoding circuit.

[0002]

2. Description of the Related Art In digital broadcasting, the form of information is M
A transport stream (hereinafter abbreviated as TS) defined by PEG2 is used, and this TS is transmitted by satellite or terrestrial wave. This TS has a packet structure of 188 bytes, and a fixed value for synchronization (47
h) is inserted as a synchronization byte. Also, in order to correct an error during transmission, a convolutional code and a Reed-Solomon code are often used in conjunction with this TS.

[0003] As a convolutional code, a punctured code using a code having a constraint length of 7 and a coding rate of 1/2 as a mother code is often used, and a Reed-Solomon code of (2
04,188). Further, between the convolutional code and the Reed-Solomon code, a byte-based interleave circuit is used for the purpose of dispersing burst errors in the convolutional decoder and improving the error correction capability.

On the other hand, when the receiving side receives the signal encoded as described above, it first convolutionally decodes the received signal with a Viterbi decoder, deinterleaves the decoding result in byte units, and Reed-Solomon decoding is performed by a Reed-Solomon decoding circuit.

[0005]

As can be seen from the above, on the digital broadcast receiving side, the problem is how to efficiently correct transmission errors and perform reliable reception.

SUMMARY OF THE INVENTION It is therefore an object of the present invention to provide an error correction system which can improve the transmission error correction efficiency and can expect reliable reception.

[0007]

SUMMARY OF THE INVENTION In order to solve the above problems, the present invention focuses on a Viterbi decoding method and a Viterbi decoding circuit based on the Viterbi decoding method. When decoding a code, a hard decision is made as to whether the received signal point is one of the expected signal points in the state transition diagram of the convolutional code, and the square of the hard decision result and the Euclidean distance between the expected signal point or It is characterized in that a branch metric to a termination state immediately before termination is obtained from an absolute value.

In the above-mentioned Viterbi decoding method and Viterbi decoding circuit, the branch metric to a state other than the termination state immediately before termination is set to an appropriate constant. In this case, an appropriate constant value is determined by the square or the absolute value of the Euclidean distance having the largest minimum distance between the signal point transitioning to the termination state immediately before termination and the signal point transitioning to a state other than the termination state immediately before termination. It is characterized by being performed.

In the above-mentioned Viterbi decoding method and Viterbi decoding circuit, a hard decision is made as to whether the received signal point is one of the signal points expected in the state transition diagram of the convolutional code, and the hard decision result and the Euclid of the expected signal point are obtained. A branch metric for the first decoding from the end state is obtained from the square or absolute value of the distance.

In the above-mentioned Viterbi decoding method and Viterbi decoding circuit, the branch metric from a state other than the termination state immediately after termination is set to an appropriate constant. In this case, the appropriate constant value is determined by the square or the absolute value of the maximum Euclidean distance between the signal point transitioning from the termination state and the signal point transitioning from a state other than the termination state. I do.

In the above-mentioned Viterbi decoding method and Viterbi decoding circuit, a branch metric is defined by replacing a received signal with a signal sequence obtained by a terminating code transition.

In the above-mentioned Viterbi decoding method and Viterbi decoding circuit, with regard to the processing during termination, the received signal is replaced with a signal sequence obtained by the code transition during termination to obtain a branch metric. A hard decision is made as to whether the received signal point is one of the expected signal points in the state transition diagram of the convolutional code, and a branch metric is calculated from the result of the hard decision and the square or absolute value of the Euclidean distance of the expected signal point. In other processes, the branch metric is obtained by soft decision.

In the above-mentioned Viterbi decoding method and Viterbi decoding circuit, the termination information of the convolutional code is obtained from a synchronization-protected output of a synchronization detection result of a transport stream defined by MPEG2 Systems.

[0014]

Embodiments of the present invention will be described below in detail with reference to the drawings.

FIG. 1 shows the configuration of a digital broadcast receiving apparatus using the Viterbi decoding method according to the present invention.
The signal captured at 1 is supplied to a tuner 2. The tuner 12 selects a signal captured by the antenna 11 and performs signal amplification and frequency conversion to an intermediate frequency. The output of the tuner 12 is supplied to a demodulation circuit 13.
The demodulation circuit 13 reproduces a clock or a carrier wave and demodulates, for example, QPSK or QAM.
The demodulated output of the demodulation circuit 13 is supplied to a Viterbi decoding circuit 14 according to the present invention.

The Viterbi decoding circuit 14 decodes a convolutionally encoded information sequence, and its decoded output is supplied to a synchronous byte detection circuit 15. The synchronization byte detection circuit 15 detects the synchronization byte (47h) added to the leading end of the decoded transport stream packet. The synchronization byte is protected by the synchronization protection circuit 16. With this synchronization protection circuit 16, the present receiver can receive stable synchronization.

The output of the Viterbi decoding circuit 14 is also supplied to a deinterleave circuit 17. This deinterleave circuit 17 disperses the residual error in the Viterbi decoded output. A Reed-Solomon decoding circuit 18 is arranged downstream of the deinterleave circuit 17. The Reed-Solomon decoding circuit 18 corrects the dispersed residual error and outputs the transport stream 9.

Here, the deinterleave circuit 17
Is intended to disperse the residual error of the Viterbi decoding circuit 14, and since a Reed-Solomon code which is a byte correction code is used outside the convolutional code,
It has a byte structure. Therefore, the synchronization protection circuit 16
The synchronization of the bit / byte conversion is achieved by the synchronization signal protected by.

Conventionally, the output of the synchronization protection circuit 16 has been supplied to the deinterleave circuit 17, but in the present invention, since the convolution code is terminated by the synchronization byte, the output of the stable synchronization protection circuit 16 is The end position is specified by using this information, and the information is also supplied to the Viterbi decoding circuit 14 so as to be used as a processing reference immediately before or during the end, or as a processing reference at the time of branching from the end state.

FIG. 2 shows an example of the internal configuration of the Viterbi decoding circuit 14. The output (I, Q) of the demodulation circuit 13 is supplied to the Viterbi decoding circuit 14, and this demodulated signal is supplied to the first and second branch metric generation circuits 141, 14
2 and two types of branch metrics are generated. The first branch metric generation circuit 141 generates a soft-decision branch metric from the received signal demodulated by the demodulation circuit 13, and the second branch metric generation circuit 142 generates a hard-decision branch metric for special processing before and after termination. What happens.

These branch metric generating circuits 14
The outputs of the selectors 142 and 142 are timing-controlled by the selector 143 based on the synchronization byte detection result at the head of the transport stream packet based on the synchronization signal protected by the synchronization protection circuit 16, and at the time of termination processing, the second branch metric generation circuit The branch metric from the first branch metric generation circuit 141 is selected from the branch metric from the first branch metric generation circuit 141 during normal processing. Using the branch metric selected in this way, ACS (Add Compare Select)
The circuit 144 calculates a state metric according to the code trellis, and uses the calculation result as a path memory circuit 145.
Is controlled, and the maximum likelihood path is traced.

The Viterbi decoding method in the Viterbi decoding circuit 14 will be described below with a specific example.

First, it is general that the bit stream transmitted by digital broadcasting conforms to MPEG Systems. However, MPEG Systems provides a special synchronization byte (47h) for TS packet synchronization as described above. ing. This synchronization byte is convolutionally encoded together with the preceding and following information parts. When the synchronization byte is regarded as a synchronization word, the length is 8 bits, so that when the constraint length of the convolutional code is 7, the convolutional code has been terminated. In the following description, for the sake of simplicity, a convolutional code having a constraint length of 3 will be described as an example, assuming that the state ends at state (0, 0).

FIG. 3 shows a convolutional coding circuit having a constraint length of 3, and FIG. 4 shows a state transition diagram of the convolutional code shown in FIG.

In FIG. 3, the convolutional coded input (x)
Are sequentially transmitted to registers 21 and 22 having a shift register configuration. The coded input (x) and the outputs of the registers 21 and 22 are added in an adder 23 to form an output "4" of modulo "2", and the coded input (x) and the register 2
The output of 2 is added by the adding circuit 24, and the modulo "2"
Of the output "6".

For the sake of simplicity, the I-axis is defined by the encoded output (y0) and the Q-axis by the encoded output (y1) of the convolutional code shown in FIG. 3, and a mapping example of QPSK modulation as shown in FIG. Think. In this example, as shown in FIG. 6, when the path ends in the state (0, 0), the following properties are provided.

(A) The state immediately before the termination is the state (0, 0) or the state (0, 1), and the state transits from either state to the state (0, 0). That is, A
Only (0,0) or C (1,1) can be taken.

(B) The branch from the termination state is the state (0,
0) or state (1, 0) only. That is, only A (0,0) or C (1,1) can be taken as a signal point.

(C) A fixed path is used during the termination period (FIG. 4).
State (0, 0)), A (0, 0) is taken as the signal point in this case.

On the other hand, as shown in FIG. 2, the Viterbi decoding circuit 14 has two blocks, an ACS circuit 144 and a path memory circuit 145 controlled by the ACS circuit 144. The ACS circuit 144 generates a branch metric defined by using the Euclidean distance between the actual reception signal point and the signal point to be received, and calculates the branch metric according to the state transition diagram with the state metric value of the state of the branch source according to the state transition diagram. The path memory circuit 145 controls the path memory circuit 145 by sequentially selecting paths having a large likelihood from two paths merging into one state, and adding the branch metric and the state metric of the selected path to a new state. Re-accumulate as metrics. This operation is repeated to sequentially perform the decoding operation.

Here, it is known from the property (a) that the modulation signal point immediately before the end is A (0,0) or C (1,1). From this, signal points A (0,0) and C
As the branch metric for (1,1), the received signal points are represented by B (1,0) and D in the mapping diagram of FIG.
By making a hard decision using a line connecting (0,1), A (0,0)
Alternatively, it is assumed that C (1, 1) has been received. in this case,
The following method can be considered as the Viterbi decoding method.

(1) The branch metric to the termination state immediately before termination is hard-determined as to which of the signal points expected in the state transition diagram of the convolutional code the received signal point is, and the hard decision result and the expected result are obtained. From the square or the absolute value of the Euclidean distance of the signal point. In this method, signal point A
It is proposed that the branch metric for (0,0) or C (1,1) be "0" or "8".

(2) The branch metric to the termination state immediately before the termination is hard-determined as to whether the received signal point is one of the signal points expected in the state transition diagram of the convolutional code. From the absolute value of the Euclidean distance of the signal point to be calculated. This method proposes that the branch metric for the signal point A (0, 0) or C (1, 1) be “0” or “2「 2 ”.

(3) An appropriate constant is set to a branch metric other than the termination state immediately before termination. In this technique,
Since it is clear from the nature of (a) that the signal points B (1,0) and D (0,1) are not taken immediately before the end, B (1,0) and D (0, It is proposed to use an appropriate value as a branch metric for 1).

(4) In (3), an appropriate constant value is
It is assumed that the minimum distance is determined by the square of the maximum Euclidean distance between a signal point transitioning to a termination state immediately before termination and a signal point transitioning to a state other than the termination state immediately before termination.
In this method, the square of the Euclidean distance between signal points B and A or B and C is used as the branch metric of signal point B (1,0), and signal points D and B are used as the branch metrics for signal point D (0,1). It has been proposed to use the square of the Euclidean distance between A or D and C. In the case of this example, the branch metrics for the signal points B (1,0) and D (0,1) are equal and "4".

(5) In (3), an appropriate constant value is
It is assumed that the minimum distance is determined by the absolute value of the maximum Euclidean distance between a signal point transitioning to a termination state immediately before termination and a signal point transitioning to a state other than the termination state immediately before termination. In this method, the absolute value of the Euclidean distance between signal points B and A or B and C as the branch metric of signal point B (1, 0), and signal points D and B as the branch metric for signal point D (0, 1). It is proposed to use the absolute value of the Euclidean distance between A or D and C. In the case of this example, the branch metrics for the signal points B (1,0) and D (0,1) are equal to “2”.

Next, it is known from the property (b) that the modulated signal point after termination is A (0,0) or C (1,1), so that the signal point A (0,0) As a branch metric to C (1,1), the received signal point is hard-decided by a line connecting B (1,0) and D (0,1) shown in FIG.
It is assumed that (0,0) or C (1,1) has been received.
In this case, the following method can be considered as the Viterbi decoding method.

(6) The branch metric for the first decoding from the termination state is hard-decided whether the received signal point is one of the signal points expected in the state transition diagram of the convolutional code, and the result of the hard decision is It is obtained from the square of the Euclidean distance of an expected signal point. In this method, the signal point A (0,
It is proposed that the branch metric for “0” or C (1, 1) be “0” or “8”.

(7) The branch metric for the first decoding from the termination state is hard-decided as to which of the signal points expected from the state transition diagram of the convolutional code the received signal point is. It is obtained from the absolute value of the Euclidean distance of the expected signal point. In this method, the signal point A (0,
It has been proposed that the branch metric for “0” or C (1,1) be “0” or “2√2”.

It is clear from the property (b) that the signal points B (1,0) and D (0,1) are not taken at the time of branching after termination. For this reason, the following method can be considered as the Viterbi decoding method.

(8) A branch metric from a state other than the termination state immediately after termination is set as an appropriate constant. This method proposes using appropriate values as branch metrics for signal points B (1,0) and D (0,1).

(9) In (8), an appropriate constant value is
The minimum distance is determined by the square of the maximum Euclidean distance between a signal point transitioning from the termination state and a signal point transitioning from a state other than the termination state. In this method, the signal point B (1,
Signal points B and A or B as branch metrics of 0)
And the square of the Euclidean distance between C and the signal point D (0,
It has been proposed to use the square of the Euclidean distance between signal points D and A or D and C as a branch metric for 1). In this example, the signal points B (1,0) and D
The branch metrics for (0,1) are equal,
It becomes "4".

(10) In (8), the minimum distance is determined by the absolute value of the maximum Euclidean distance between the signal point transitioning from the termination state and the signal point transitioning from a state other than the termination state. In this method, the absolute value of the Euclidean distance between signal points B and A or B and C as the branch metric of signal point B (1, 0), and signal points D and B as the branch metric for signal point D (0, 1). It is proposed to use the absolute value of the Euclidean distance between A or D and C. In this example, the signal points B (1,0) and D (0,
The branch metrics for 1) are equal and equal to "2".

Furthermore, because of the nature of (c), during the termination period, a fixed path is followed (state (0, 0) in FIG. 4). In this case, the original transmission is performed regardless of the reception signal point. The signal is known to be A (0,0). For this reason, the following method can be considered as the Viterbi decoding method.

(11) For the code transition during termination, the branch metric is defined by replacing the received signal with a signal sequence obtained by this code transition. In this technique,
Each signal point A (0,0), B (1,0), C (1,1),
It has been proposed to use the square or absolute value of the Euclidean distance between each signal point and A (0,0) as a branch metric for D (0,1).

(12) Regarding the processing during termination (11)
The metric is generated by one of the hard decision processes (1), (2), (6), and (7) immediately before and immediately after the termination. Is used. This method proposes that a soft decision is made in a portion other than the termination process, and a hard decision branch metric (1) to (11) is adopted in the termination process.

(13) The termination information of the convolutional code is obtained from an output in which the synchronization detection result of the transport stream is synchronized and protected. In this method, a synchronization byte (47h) of a transport packet is detected, a synchronization protection is applied to the detection result, a stable synchronization is supplied, and the end position is accurately determined by the stable synchronization. It is proposed to perform the termination processing of (12).

As described above, in the conventional Viterbi decoding method, processing is performed without using the condition that the convolutionally coded bit sequence is terminated at the synchronization byte part. In the method, the synchronization part transmitted periodically is detected, the end part of the code is known using the synchronization signal subjected to high reliability processing, and Viterbi decoding is performed in consideration of the termination condition of this code. Therefore, it is possible to obtain characteristics higher than those of the related art.

In particular, MPEG systems are generally adopted in digital broadcasting, and in this case, a transport stream is used. At this time, a 1-byte synchronization signal is added to the head of the packet, and when the signal is convolutionally coded, the synchronization byte performs the termination operation of the convolutional code. Therefore, by appropriately processing the termination portion, the transmission characteristics can be improved.

[0050]

As described above, according to the present invention, it is possible to improve the efficiency of correcting transmission errors and to provide an error correction system that can be expected to have reliable reception.

[Brief description of the drawings]

FIG. 1 is a block circuit diagram showing a configuration of a digital broadcast receiving apparatus using a Viterbi decoding method according to the present invention as an embodiment of the present invention.

FIG. 2 is a block circuit diagram showing a specific configuration of the Viterbi decoding circuit according to the embodiment.

FIG. 3 is a block circuit diagram showing a specific example of a convolutional encoding circuit (constraint length 3) used on the transmitting side of digital broadcasting.

FIG. 4 is a trellis diagram showing a state transition of the convolutional coding circuit (constraint length 3) shown in FIG. 3;

FIG. 5 is a QPS as an example of a digital broadcasting modulation method.
The figure which shows the example of a mapping of K.

FIG. 6 is a trellis diagram for explaining path merging at the time of the termination operation of the embodiment.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 11 ... Antenna 12 ... Tuner 13 ... Demodulation circuit 14 ... Viterbi decoding circuit 141 ... 1st branch metric generation circuit 142 ... 2nd branch metric generation circuit 143 ... Selector 144 ... ACS circuit 145 ... Path memory circuit 15 ... Synchronous byte detection circuit 16 ... Synchronization protection circuit 17 ... Dinder leave circuit 18 ... Reed-Solomon decoding circuit 21,22 ... Register 23,24 ... Addition circuit

Claims (14)

[Claims] 1. A Viterbi decoding method for decoding a convolutional code subjected to a termination process, which transits to a fixed position in a fixed pattern, wherein a received signal point is set to one of signal points expected in a state transition diagram of the convolutional code. A Viterbi decoding method characterized in that a hard decision is made as to whether or not there is, and a branch metric to a termination state immediately before termination is obtained from the square or the absolute value of the Euclidean distance of an expected signal point and the result of the hard decision. 2. A Viterbi decoding method for decoding a convolutional code subjected to termination processing, which transitions to a fixed position in a fixed pattern, wherein a branch metric other than the termination state immediately before termination is set to an appropriate constant value, and Is characterized in that a minimum distance is determined by a square or an absolute value of a maximum Euclidean distance between a signal point transitioning to a termination state immediately before termination and a signal point transitioning to a termination state other than the termination state immediately before termination. Viterbi decoding method. 3. A Viterbi decoding method for decoding a convolutional code which has been subjected to a termination process and transiting to a fixed position in a fixed pattern, wherein a received signal point is set at one of signal points expected in a state transition diagram of the convolutional code. A Viterbi decoding method characterized in that a hard decision is made as to whether or not there is a branch metric for the first decoding from a termination state from the result of the hard decision and the square or absolute value of the Euclidean distance of an expected signal point. 4. A Viterbi decoding method for decoding a convolutional code subjected to termination processing, which transits to a fixed position in a fixed pattern, wherein a branch metric from a state other than the termination state immediately after termination is set to an appropriate constant value, Is a Viterbi decoding method in which a minimum distance is determined by a square or an absolute value of a maximum Euclidean distance between a signal point transitioning from a termination state and a signal point transitioning from a state other than the termination state. 5. A Viterbi decoding method for decoding a convolutional code subjected to termination processing, which transits to a fixed position in a fixed pattern, wherein a branch metric is defined by replacing a received signal with a signal sequence obtained by a transition of the code being terminated. A Viterbi decoding method. 6. A Viterbi decoding method for decoding a convolutional code that has been subjected to a termination process, which transitions to a fixed position in a fixed pattern, wherein the process during the termination is a signal sequence obtained by a code transition during termination of a received signal. In the processing immediately before and immediately after the termination, the hard decision is made as to whether the received signal point is one of the signal points expected in the state transition diagram of the convolutional code, and the result of the hard decision and A Viterbi decoding method characterized in that a branch metric is obtained from a square or an absolute value of a Euclidean distance of an expected signal point, and in other processes, a branch metric is obtained by soft decision. 7. The termination information of the convolutional code is MPE
7. The Viterbi decoding method according to claim 1, wherein a synchronization detection result of a transport stream defined by G2 Systems is obtained from a synchronization-protected output. 8. A Viterbi decoding circuit for decoding a convolutional code subjected to termination processing, which transits to a fixed position in a fixed pattern, wherein a received signal point is set to one of signal points expected in a state transition diagram of the convolutional code. A Viterbi decoding circuit characterized in that a hard decision is made as to whether or not there is, and a branch metric to a termination state immediately before termination is obtained from the square or the absolute value of the Euclidean distance of the expected signal point and the expected signal point. 9. A Viterbi decoding circuit for decoding a convolutional code subjected to termination processing, which transits to a fixed position in a fixed pattern, wherein a branch metric other than the termination state immediately before termination is an appropriate constant value, Is characterized in that a minimum distance is determined by a square or an absolute value of a maximum Euclidean distance between a signal point transitioning to a termination state immediately before termination and a signal point transitioning to a termination state other than the termination state immediately before termination. Viterbi decoding circuit. 10. A Viterbi decoding circuit for decoding a convolutional code subjected to termination processing, which transits to a fixed position in a fixed pattern, wherein a received signal point is set to one of signal points expected in a state transition diagram of the convolutional code. A Viterbi decoding circuit characterized by hard-deciding whether or not there is, and obtaining a branch metric for the first decoding from a termination state from the result of the hard decision and the square or absolute value of the Euclidean distance of an expected signal point. 11. A Viterbi decoding circuit for decoding a convolutional code subjected to termination processing which transits to a fixed position in a fixed pattern, wherein a branch metric from a state other than the termination state immediately after termination is set to an appropriate constant value, A Viterbi decoding circuit characterized in that a minimum distance is determined by a square or an absolute value of a maximum Euclidean distance between a signal point transitioning from a termination state and a signal point transitioning from a state other than the termination state. 12. A Viterbi decoding circuit for decoding a convolutional code subjected to termination processing, which transits to a fixed position with a fixed pattern, wherein a branch metric is defined by replacing a received signal with a signal sequence obtained by a transition of the code being terminated. A Viterbi decoding circuit. 13. A Viterbi decoding circuit for decoding a convolutional code subjected to a termination process that transits to a fixed position in a fixed pattern, wherein a process during termination is a signal sequence obtained by a code transition during termination of a received signal. In the processing immediately before and immediately after the termination, the hard decision is made as to whether the received signal point is one of the signal points expected in the state transition diagram of the convolutional code, and the result of the hard decision and A Viterbi decoding circuit characterized in that a branch metric is obtained from a square or an absolute value of an Euclidean distance of an expected signal point, and in other processing, a branch metric is obtained by soft decision. 14. The termination information of the convolutional code is MP
14. The Viterbi decoding circuit according to claim 8, wherein a synchronization detection result of a transport stream defined by EG2 Systems is obtained from a synchronization-protected output.