KR20010037341A - Puncturing pattern search method for convolution code - Google Patents

Abstract

PURPOSE: A method for searching a puncturing pattern of a convolution code is provided to classify an input bit sequence into three parts, determine a wait distribution and find an optimal puncturing table using the determined result. CONSTITUTION: In a method for searching a puncturing pattern of a convolution code, an input bit sequence is consisted of a first interval having an error path by a bit value "1", a tail bit interval formed by "0" sequence making a zero state and an optional sequence interval located between the first interval and the tail bit interval. An input sequence and a puncturing pattern are loaded and a coding process is performed, while a wait distribution for the input bit sequence is determined. The determined wait value is applied in a particular formula and thereby an optimal puncturing pattern is obtained.

Description

Punching pattern search method of convolution code {PUNCTURING PATTERN SEARCH METHOD FOR CONVOLUTION CODE}

BACKGROUND OF THE INVENTION 1. Field of the Invention [0001] The present invention relates to a technique for properly changing a channel environment in a wireless mobile communication system or to appropriately adjusting a data rate during multimedia data transmission. The present invention relates to a puncturing pattern search method of a convolutional code that is suitable for providing an optimal puncturing pattern used to obtain various data rates when using a code.

Recently, various techniques for improving the reliability of information in a wireless mobile communication system have been developed. Among them, the convolution coding method adopting the Viterbi algorithm as one of the error correction techniques has been applied to many systems. For example, the coding scheme is applied to and used in mobile communication systems such as IS-95 and GSM, and has already been adopted in mobile communication systems (eg, IMT-2000) that will be introduced in the future. The reason why the convolutional coding scheme is widely used is that it has the advantage of obtaining a high code gain even when using a relatively simple decoding algorithm compared to other coding schemes.

When the environment of the channel is flexibly changed or the multimedia data is to be transmitted, the data transmission efficiency can be improved by appropriately changing the transmission side rather than sticking to a fixed data rate.

An example of a method of changing the data rate is a method of designing an optimal code end and a decoding end suitable for each data rate, but in this case, it is necessary to include both the code end and the decoding end corresponding to each data rate. There is a problem that the system is too complicated.

Another method of changing the data rate is to use a puncturing table, which can be used to obtain a variety of data rates with one code and a decoding end having a low data rate. In this case, since the performance varies greatly depending on the puncturing pattern, the optimal puncturing table is an important parameter of the convolution code that changes the data rate.

Currently, a table is known for mapping a convolutional code having a fixed data rate of 1/2 to various data rates, but there is no known technique for puncturing patterns for convolutional codes having a data rate of 1/3. Since convolutional codes of length 9 and data rate 1/3 are adopted in various channels of next-generation mobile communication, puncturing patterns become an important issue when changing data rates for flexible data services.

However, in the conventional Viterbi decoder puncturing pattern construction method according to the computer simulation, the weight distribution is determined according to a normal input bit string input to the decoder, and the puncturing table is constructed using the weight distribution pattern. Difficulties in maintaining optimal performance rates.

Accordingly, an object of the present invention is to determine the weight distribution by dividing the input bit string into three parts in consideration of the fact that the input bit string is punctured and punctured by a puncturing table having a puncturing period. The present invention provides a method for searching a puncturing pattern to find a puncturing table.

Figure 1a is a block diagram of a transmitting end applied to the present invention.

Figure 1b is a block diagram of the receiving end applied to the present invention

2 is an exemplary block diagram of an encoder in FIG. 1A;

3 is a signal flowchart of an input bit string generation method according to the present invention;

4 is a diagram illustrating classification of input bit strings according to the present invention;

5 is a signal flowchart of a weight distribution generation method according to the present invention;

6 is a bit error rate graph for an optimal puncturing table.

*** Description of the symbols for the main parts of the drawings ***

11A: Encoder 11B: Punching Table

12A: Depunching Table 12B: Decoder

In order to achieve the object of the present invention, a convolutional code puncturing pattern search method includes a section having an incorrect path by a bit value ″ 1 ″, a tail bit section formed by a ″ 0 ″ sequence which makes the state zero. A first step of generating an input bit string consisting of an arbitrary sequence section located in the middle of the section; A second process of determining a weight distribution for the input bit string while performing an encoding process by loading an input sequence and a puncturing pattern; The third step is to find an optimal puncturing pattern by applying the weight value obtained by the second step to a specific equation.

Fig. 1A is a block diagram of a transmitter according to the present invention, as shown therein, which encodes a series of bit streams and outputs them in a different form from the original bit stream; When transmitting the encoded bit stream output from the encoder 11A to the channel, the puncturing table 11B is configured to remove the convolution code according to the puncturing pattern in order to reduce the loss of the bandwidth.

1B is a block diagram of a receiver applied to the present invention, as shown therein, a depuncturing table 12A for inserting data into a punctured portion for data received through a channel after being sent from the transmitter; It is composed of a decoder 12B which recovers the original bit string from the bit string compensated by the depuncturing table 12A, and is described in detail with reference to FIGS. 2 to 6 attached to the operation of the present invention. The explanation is as follows.

The amount of data output through the convolutional encoder 11A becomes larger than the data amount to be originally transmitted, because the output data has history information of data that is continuous in a different form from the original bit stream. to be. Low-late convolutional code has quantitative redundancy than source code.

This allows the error-free code to be decoded in the Viterbi decoder when decoding to source code, but in terms of coding gain, the bandwidth is large, resulting in significant losses. Therefore, the puncturing operation is performed through the puncturing table 11B in order to reduce the loss of bandwidth (coding gain loss) caused thereby.

In this case, when the change target code is a punctured convolutional code, the data rate is adjusted using a puncturing pattern having a period P based on a convolution code having a fixed low data rate (1 / N). The data rate r converted in this way is expressed by the following expression (1).

---------------- (Eq. 1)

At this time, the puncturing pattern is expressed by the following expression (2).

If is, the encoded signal is transmitted through the channel as it is. In the case of, since the coded signal is discarded, it is not transmitted through the channel.

Meanwhile, the receiving end needs to restore data lost due to the puncturing, that is, insert data into the punctured portion before the data is input to the Viterbi decoder 12B, which is a depuncturing table. Performed by 12A.

When the decoding stage knows the puncturing table 12A of the code in advance, a dummy signal is inserted into the signal removed from the code stage to use a decoding algorithm corresponding to the mother code 1 / N. At this time, the decoding algorithm is applicable to the Viterbi algorithm used when the length of the constraint field is small and the sequential algorithm used when the length of the constraint field is relatively long, the present invention adopts the Viterbi algorithm that can calculate the error rate mathematically It was.

The convolutional code applied to the embodiment of the present invention has a restriction length of 9 and a data rate of 1/3. And the generated polynomials that produce the encoded data , , to be.

As a result, the bits {X (1), X (2), X (3)} encoded through the encoder 11A as shown in FIG. 2 are delivered to the puncturing table 11B, and the convolutional code according to the puncturing pattern thereof. Is removed and then transmitted through the channel, and the signal transmitted through the channel is then subjected to a decoding algorithm after the bits removed by the depuncturing table 12A are compensated again.

Hereinafter, a process of obtaining an optimal puncturing pattern according to the present invention will be described in detail.

The present invention proposes a puncturing pattern that minimizes the theoretical bit error rate in order to obtain an optimal puncturing pattern. When decoding the convolution code using the soft decision Viterbi algorithm, the upper bound of the bit error rate can be obtained by using the weight distribution of the encoded bit string, which is expressed by the following equation (5).

---------------------- (Eq. 5)

At this time, Is the minimum distance of the coded bit stream, Is the number of cases where the weight of the coded bit string is d when the wrong path is selected in the Viterbi algorithm. The error rate for one bit in a Gaussian channel is represented by the following equation (6), and if the wrong bits in the coded bit stream do not affect each other, Can be expressed by the following equation (7).

------------------------- (Equation 6)

Where E is the energy of the transmitted signal, Means noise variance.

------------------------- (Eq. 7)

Using the above equation, the weight distribution of the coded bit string corresponding to the puncturing pattern in all cases can be obtained and the upper limit of the bit error rate You can find a puncturing pattern that minimizes

A method of generating a bit string to be input to the encoder 11A will now be described with reference to FIG. 3.

The zero bit ″ 0 ″ is continuously inserted into the input string until the preset integer value I reaches the puncturing period, and the bit value ″ 1 ″ is inserted when the integer value I reaches the puncturing period (SA1-). SA5). Subsequently, random bit values are inserted over a predetermined number of times, and tail bit values are inserted over a predetermined number of times (SA6 and SA7).

Since the input bit string is punctured by a puncturing table having a puncturing period, it is necessary to generate an input bit string according to the following algorithm, which is divided into three sections as shown in FIG. 4.

The first section is a section having a wrong path by the bit value ″ 1 ″. And a tail bit section formed by a ″ 0 ″ sequence that makes zero, and finally, an arbitrary sequence section located in the middle of the two sections.

The weight distribution according to each input is determined according to the algorithm of FIG. 5 using the input bit string as described above, and the resultant value (weight value) is applied to Equation 5 to find an optimal puncturing table.

That is, the weight distribution generation process loads the generated input sequence and loads the puncturing pattern as shown in FIG. 5 (SB1 and SB2). Thereafter, the weight distribution output from the encoder 11A is stored (SB3 and SB4).

A bit error rate graph for the optimal puncturing table 11B is shown in FIG. 6. That is, FIG. 6 shows the theoretical upper limit error rate according to Equation 5 when the signal-to-noise ratio S / N is 5 dB and the puncturing periods are 2, 3, 4, 5, and 6, respectively.

As a result, in the present invention, computer simulation was used in the puncturing pattern search process as described above. The results are as follows.

As described in detail above, the present invention requires a variety of data rates in order to provide a multimedia service in a communication system using a constriction length of 9 and using a 1/3 convolution code, or continuously recursively adjusts channel states. In the structure of the transmitter that obtains the information, the input bit string is divided into three parts to determine the weight distribution, and then the optimal puncturing table can be found by using the result, thereby constructing the optimal puncturing table. The transmission performance is improved as much as possible.

Claims (5)

A first step of generating an input bit string comprising a section having a wrong path by a bit value ″ 1 ″, a tail bit section made by a ″ 0 ″ sequence, and an arbitrary sequence section located between the two sections; Determining a weight distribution for the input bit stream; And a third process of deriving an optimal puncturing pattern by applying the weight value obtained by the second process to the following equation. : Upper limit of bit error rate : Minimum distance of coded bit stream The number of cases where the weight of the coded bit string is d when the wrong path is selected in the Viterbi algorithm. The method of claim 1, wherein the first process continuously inserts a zero bit ″ 0 ″ into the input string until a predetermined integer value reaches a puncturing period, and when the integer value I reaches the puncturing period, the bit value ″ 1. Inserting " And a second step of inserting a random bit value over a predetermined number of times and then inserting the tail bit value over a predetermined number of times. The method of claim 1, wherein the second process comprises: a first step of loading the generated input sequence and puncturing pattern; And a second step of storing a weight distribution outputted from the encoder. The method of claim 1, wherein the puncturing pattern is applied to a wired / wireless communication system using a k = 9, 1/3 convolutional code of constraint length. The method of claim 1 or 4, wherein the puncturing pattern is applied when the puncturing period is 2,3,4,5,6.