JPH06164424A - Successive decoding device - Google Patents

Abstract

PURPOSE:To effectively prevent erroneous correction, which concentratedly occurs before and after overflow, of a sequence encoded to a convolutional code by controlling operation and stop of error correction in accordance with positional relations between the pointer of a delay buffer and that of a trace buffer. CONSTITUTION:A decoder is provided with an erroneous correction preventing circuit 105 which compares a pointer 1 of the delay buffer where a reception sequence is stored with a pointer 2 of the trace buffer where the results of maximum likelihood path search are stored in the time series order. The pointer 1 points the address of the oldest data in the reception sequence stored in the delay buffer, and the pointer 2 points the front end node of the searched maximum likelihood path. When the difference between pointers 1 and 2 is equal to or smaller than a certain value, the error correcting operation is stopped by a control signal 5 from the error correction preventing circuit 105. When the difference between pointers 1 and 2 is larger than the certain value, the error correcting operation is started by the control signal from the erroneous correction preventing circuit 105.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a successive decoding error correction device which operates in a continuous mode.

[0002]

2. Description of the Related Art Error correction techniques are widely used in order to provide high quality broadcasting, communication lines or recording devices. Taking a digital communication line as an example, the information sequence to be sent is coded so that it has redundancy and then transmitted. When an error occurs in the information sequence, the receiving side uses the redundancy to some extent. You can correct even errors. There are many encoding and decoding methods, among which is the convolutional coding / sequential decoding method. What is the sequential decoding method?
By using the tree structure of the convolutional code and searching for the most probable branch (called the maximum likelihood path) at that time,
It is one decoding means for judging the transmitted information sequence.

The successive decoding method has a feature that the time required for decoding varies. That is, the number of calculations required to search the maximum likelihood path is not always constant but given as an average value. In the successive decoding device that operates in the continuous mode, the time required for decoding is absorbed by the delay buffer, but the number of calculations is limited by the size of the delay buffer. When the maximum likelihood path cannot be searched within the limited number of calculations, the sequential decoding device cannot perform error correction. This event is called overflow and occurs with a certain probability.

An outline of the operation of the conventional successive decoding apparatus will be described. FIG. 2 is a block diagram showing the configuration of the successive decoding apparatus. The input / output unit 102 receives the encoded signal as the reception sequence 14, stores the received signal in the delay buffer 100 through the address line 10 and the data line 12, and also receives the delay buffer 100 in response to the request from the error estimation unit 103.
The received data at the specific time stored in the above is transferred to the error estimation unit 103 by the received signal 17. Error estimation unit 103
Searches the maximum likelihood path from the received data received from the input / output unit 102, and stores the result in the trace buffer 101 through the address line 11 and the data line 13.

The address at which the oldest received data is stored in the delay buffer 100 is the pointer 1
Are managed in the input / output unit 102. When an input request for a new reception sequence 14 arrives, the input / output unit 102 transfers the reception data at the address of the pointer 1 to the delay buffer 10.
The correction signal 16 corresponding to the pointer 1 is fetched from 0, the error estimation unit 103 is requested to perform the correction signal 16 and the corrected signal 16 is output to the output signal 15. Next, the reception sequence 14 input to the input / output unit 102 is stored at the address of the pointer 1 of the delay buffer 100, and the pointer 1 is updated by increment of 1.

On the other hand, the trace buffer 101 stores the maximum likelihood path at that time. The error estimation unit 103 manages the depth of the node at the tip of the nodes forming this path as the pointer 2. When the number of errors occurring on the communication path is small, the number of calculations required to search the maximum likelihood path is small. Therefore, the pointer 2 is updated by increment of 1 each time the reception sequence 14 is input, and the pointer 2 is delayed as compared with the pointer 1. In many cases, they are separated by the size of the buffer 100.

However, under high noise, the number of calculations for searching the maximum likelihood path increases exponentially, so that it becomes difficult for the pointer 2 to find a correct path, and the pointer 2 becomes stagnant. As a result, the pointer 1 may catch up with the pointer 2. The control unit 104 determines that this is an overflow,
An initialization command is issued to the error estimation unit 103 through the control signal 4, and error correction is not performed during that time. The error correction operation is restarted immediately after the initialization operation is completed.

[0008]

In a channel with high noise such as overflow, the maximum likelihood path at a certain point can be searched within a predetermined number of calculations, but the reliability is low. Often the wrong path. As a result, erroneous correction (called erroneous correction) deteriorates the error rate after decoding. Such phenomena occur intensively before and after the overflow occurs. In the conventional technique, the error correction operation is continued until immediately before the overflow occurs, and the error correction operation is started immediately when the post-processing at the time of overflow is completed and the error correction operation can be restarted.

The present invention provides a sequential decoding device capable of effectively preventing erroneous corrections that occur concentratedly before and after an overflow in a convolutionally encoded sequence.

[0010]

In order to achieve this object, in a sequential decoding device operating in a continuous mode, a pointer of a delay buffer for storing a received sequence and a pointer of a trace buffer for storing a searched maximum likelihood path are provided. It is provided with a circuit for comparison and a switch capable of switching the operation / stop of the error correction operation, and the operation and stop of the error correction operation can be controlled by the positional relationship between the pointers.

[0011]

The decoder is provided with an erroneous correction prevention circuit for comparing the pointer 1 of the delay buffer for storing the reception sequence with the pointer 2 of the trace buffer for storing the result of searching the maximum likelihood path in chronological order. The pointer 1 indicates the address of the oldest data in the reception sequence stored in the delay buffer, and the pointer 2 indicates the leading node of the searched maximum likelihood path. When the difference between the pointer 1 and the pointer 2 is within a certain value A, the control signal 5 from the error correction prevention circuit
Stops the error correction operation. When the difference between the pointer 1 and the pointer 2 is more than a certain value B, the error correction operation is started by the control signal 5 from the error correction prevention circuit. This prevents erroneous corrections that often occur before and after the overflow peculiar to the sequential decoding device, and improves the error rate characteristic after decoding.

[0012]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS FIG. 1 shows the device configuration of the present invention. Threshold A
In the erroneous correction prevention circuit 105 in which the threshold value B and the value of the pointer 2 are input,
How close pointer 1 is to pointer 2 is determined from the difference (called the difference between the two pointers). If the difference between both pointers becomes smaller than the threshold value A while the error correction operation is operating, the error correction prevention circuit 105
Command the stop of the error correction operation, which is transmitted to the switch 106 by the control signal 5. The control signal 5 is input to the switch 106, which has a function of controlling the operation / stop of the error correction operation. That is, the ground 19 is output as the correction signal 18 when the error correction operation is stopped, and the correction signal 16 is output when the error correction operation is activated.
Further, when the difference between the two pointers exceeds the threshold value B while the error correction operation is stopped, the error correction prevention circuit commands the operation of the error correction operation and is transmitted to the switch 106 through the control signal 5. At this time, the switch 106 outputs the correction signal 16 as the correction signal 18.

Here, the optimum values of the threshold value A and the threshold value B differ depending on the coding rate in the encoder, the constraint length, the state of the communication path, and the like. In general, if the threshold value A or the threshold value B is set too large, the error correction capability of the iterative decoding apparatus will be reduced. By properly setting the threshold value A and the threshold value B according to the encoder used and the state of the assumed communication channel, after decoding, especially in a high noise communication channel in which overflow frequently occurs. The bit error rate characteristic of is greatly improved.

[0014]

As described above, according to the present invention,
Error correction can be performed on convolutionally coded sequences that continue infinitely, and it is possible to effectively prevent erroneous corrections that are concentrated before and after the overflow that occurs in conventional sequential decoding devices. The error rate performance of can be improved.

[Brief description of drawings]

FIG. 1 is a block diagram showing an embodiment of the present invention.

FIG. 2 is a block diagram showing an example of a sequential decoding device configured by a conventional technique.

[Explanation of symbols]

 1 pointer 1 2 pointer 2 3 control signal 3 4 control signal 4 5 control signal 5 10 address line 11 address line 12 data line 13 data line 14 reception sequence 15 output signal 16 correction signal 17 reception signal 18 correction signal 19 ground 100 delay buffer 101 trace buffer 102 input / output unit 103 error estimation unit 104 control unit 105 erroneous correction prevention circuit 106 switch

Claims (2)

[Claims] 1. A circuit for comparing a pointer of a delay buffer storing a received sequence with a pointer of a trace buffer storing a searched maximum likelihood path in a successive decoding apparatus operating in a continuous mode, and activation / stopping of an error correction operation. And a switch capable of switching between the two pointers, and the successive decoding apparatus is configured to control the operation and stop of the error correction operation according to the positional relationship between the pointers. 2. The error correction operation is stopped when the difference between the two pointers is within a threshold value A, and the error correction operation is performed when the difference between the pointers is more than a threshold value B, and the threshold values A and B are A ≦
It is B, The sequential decoding apparatus of Claim 1 characterized by the above-mentioned.