KR20050079986A - Apparatus and method for controlling iterative decoding for turbo decoder using compare of llr's sign bit - Google Patents

Abstract

The present invention uses LLR code comparison to reduce the decoding delay by comparing and counting the code bits of the LLR, which is a soft decision output value of a serial turbo decoder, to stop after a predetermined iteration using a different degree of code bits. It relates to an iterative decoding control device and method of a turbo decoder. An iterative decoding control apparatus of a turbo decoder using a sign comparison of an LLR is a turbo decoder, and includes a soft decision value output means for repeatedly decoding a turbo-coded data and outputting a soft decision value, and a soft decision value output from the soft decision value output means. By comparing the codes, counting the cases where the signs are different, or when the number of repeated decoding exceeds the set maximum number of repeated decoding, it is determined whether the repeated decoding control means for stopping the repeated decoding of the turbo decoder and whether the repeated decoding control means performs the repeated decoding. And hard decision value output means for hardly determining the output value of the soft decision value output means and outputting the decrypted result.

Description

Apparatus and method for controlling iterative decoding for turbo decoder using compare of LLR's sign bit}

The present invention relates to an iterative decoding controller of a turbo decoder in a wireless communication system and a method thereof, and more particularly, to compare and count the code bits of the LLR, which is a soft decision output value of a turbo decoder configured in series, to use different degrees of code bits. The present invention relates to an iterative decoding control apparatus and method of a turbo decoder using a comparison of codes of an LLR which can reduce a decoding delay by stopping after a predetermined iterative decoding.

In general, in the field of wireless communication such as mobile communication, many errors occur due to complex phenomena such as noise and interference generated on a wireless channel, compared to wired communication. Robust error correction codes are used to correct errors occurring in the wireless channel. The turbo encoder used as an error correction code is concatenated and encoded recursive systematic convolutional code (RSC) in parallel with an interleaver, and is performed through iterative decoding and quasi-optimal decoding. In addition, the turbo encoder shows an excellent performance approaching Shannon's Limit in terms of bit error ratio (BER) when the interleaver is large in size and repeated decoding is sufficiently performed.

1 is a block diagram of an encoder for encoding a turbo code at a transmitting end in a general wireless communication system, and passes through the input information bit d _k as it is and outputs it as one information signal X _k (100). A second encoder for encoding the interleaver 103 and the interleaver information bits 104 which reduce correlation by different input orders of the first encoder 101 for encoding (d _k ) and the input information bits d _k . It consists of 105.

Accordingly, the turbo encoder outputs dual parity bits of y _1k 102 and y _2k 106, which are outputs of the original information signal X _k 100, the first encoder 101, and the second encoder 102. Done.

Also, in order to obtain a desired coding rate in the turbo encoder, the first encoder 101 and the second encoder 105 are punctured by the output signal Y _k through the puncturer 107. In this way, it is transmitted to the receiver through the channel together with the encoder output finally obtained and the original information signal X _k (100).

To decode a codeword coded by the turbo encoder of FIG. 1 at a receiving end, two decoders having the structure as shown in FIG. 2 are connected in series and decoded. The decoder uses the Maximum a Posteriori (MAP) and Soft Output Viterbi Algorithm (SOVA) decoder, and the MAP decoder is known to have better general BER performance.

Unlike the conventional multi-level coders such as Concatenated Codes, the turbo coder can perform the decoding operation repeatedly by exchanging external information between the two decoders.

2 is a block diagram showing the configuration of a conventional turbo decoder. Hereinafter, a decoding process of a general turbo decoder will be described with reference to the turbo decoder of FIG. 2.

First, the first decoder 204 performs a decoding process using the information bit X _k 200 and the parity bit y _1k 202 received through the channel and outputs a decoded result. The interleaver 206 performs an interleaving process to make the burst error present in the output signal 205 of the first decoder 204 into a random error, and the interleaved signal 207. ) The second decoder 208 performs the decoding process again using the interleaved signal 207 and the parity bit y _2k 203 and outputs a decoded result.

At this time, if iterative decoding is not desired, the second deinterleaver 210 deinterleaves the output signal 209 of the second decoder 208, and the hard determiner 212 deinterleaves the signal 211. Decision signal by hard decision Output to (213). However, in order to perform repeated decoding, the external information value is extracted from the output signal of the second decoder 208 and fed back to the first decoder 204 through the first deinterleaver 215. This feedback signal is used as the pre-information value 216 in the first decoder 204 to play a role of improving the reliability of the information bits during iterative decoding. This iterative decoding is continued until the desired performance is achieved.

Conventionally, in the wireless communication field such as mobile communication, many transmission errors occur due to channel noise due to propagation delay and fading due to reception of multiple radio paths. Error correction codes used to improve these problems and to increase the reliability of data have become an important factor in digital mobile communication systems.

Recently announced turbo codes are encoded by connecting RSC codes in parallel and performing decoding operations through iterative decoding. In addition, the turbo code shows an excellent performance approaching the Shannon limit in terms of BER when the interleaver is large in size and repeated decoding is sufficiently performed.

However, the conventional methods have various problems in terms of complexity increase due to a large amount of computation, decoding delay due to interleaver and iterative decoding, and real time processing.

 The turbo decoder, which is a decoding device of a turbo code, performs decoding according to a predetermined number of iterations before input. The hard-decision of the LLR value, which is the final output obtained after an arbitrary number of repetitive decoding times, is assumed as a threshold to estimate the input information bit value of the turbo encoder.

In this case, as the number of iterations increases, the BER gets better, but the delay time of data transmission is limited. In the turbo coder, even if the number of iterations increases, the coding gain no longer improves. do.

In addition, an increase in the number of iteration decoding for the improvement of the BER performance causes an increase in the decoding time, which causes a problem that the number of iteration decoding cannot be continuously increased in order to obtain a desired BER performance in an actual system. Therefore, this iterative decoding technique is inefficient in turbo decoders having a certain level of performance and has a problem of delaying data decoding.

In order to solve this problem, a stop criterion is required to stop decoding after a predetermined iterative decoding. Until recently, the Stop Criterion by Cross Entropy (Hagenauer) proposed by Hagenauer has been used a lot, but this method includes more complicated calculations, which leads to increased complexity due to a large amount of computation and There are many problems such as difficulty in hardware implementation.

The technical problem to be achieved by the present invention is to reduce the decoding delay by comparing and counting the code bits of the LLR, which is a soft decision output value of a serial turbo decoder, and stopping them after a predetermined iteration using different degrees of code bits. It is an object of the present invention to provide an iterative decoding control device and method for a turbo decoder using a sign comparison of LLR.

An iterative decoding control apparatus of a turbo decoder using a code comparison of an LLR for solving the above technical problem to be solved by the present invention is a turbo decoder, and soft decision value output means for outputting a soft decision value by repeatedly decoding turbo encoded data. ; Repetitive interruption of repeating the decoding of the turbo decoder by counting the case where the sign is different by comparing the codes of the soft decision values output from the soft decision value output means or when the number of repeated decoding exceeds the set maximum number of repeated decoding. Control means; And hard decision value output means for hardly determining the output value of the soft decision value output means and outputting the decoded result according to whether or not the iteration decoding control means repeatedly performs the decoding.

In the present invention, a maximum repetition number of decoding is set, and further comprising a counter for counting the number of repetitions of decoding from the soft decision value of the soft decision value output means and outputting the result to the repetition stop control means. .

In the present invention, the soft decision value output means comprises: a first soft decision unit for soft decision of the first decoded data; And a second soft decision unit for soft decision of the second decoded and interleaved data.

In the present invention, the repeat stop control means compares the signs of the soft decision values output from the soft decision value output means to count the case where the signs are the same, or when the number of iteration decoding is less than the set maximum iteration decoding number, the turbo It characterized in that the control is performed so that repeated decoding of the decoder continues.

In the present invention, the repeat stop control means includes: a first code storage unit for storing a code of a bit string output from the first soft determiner; A second code storage unit which stores a code of a bit string output from the second soft determiner; A comparator for comparing codes of the first code storage unit and the second code storage unit; A down counter for counting down count values set in descending order when the codes of the first code storage unit and the second code storage unit are different as a result of the comparison; A counter for counting the number of times of repeated decoding; And a repeat decoding stop signal generator for outputting a repeat decoding stop signal to the hard decision output means when the down counter counts a non-zero value or when the counter performs a set maximum number of repetitions. .

The iterative decoding control method of the turbo decoder using the comparison of the code of the LLR for solving the technical problem to be achieved by the present invention is a iterative stop control method applied to the turbo decoder, (a) the soft decision of the first decoded data Storing the code; (b) storing the sign of the soft-determined second decoded and deinterleaved data; (c) performing down counting of the down counters by comparing the codes and counting the cases of different codes in descending order; (d) comparing a value of counting the number of times of repetition decoding by a counter with a maximum number of times of repetition decoding previously set in the counter; (e) generating a repeat decoding stop signal when the count value of the down counter is not 0 or the counter exceeds the maximum number of repeated decoding; And (f) hard determining and outputting the soft decoded second decoded and deinterleaved data.

In the present invention, when the down counter counts the same code or when the counter counts less than the maximum number of repeated decoding for which the number of repeated decoding is set, it is characterized in that the repeated decoding of the turbo decoder is continuously performed.

In the present invention, when iterative decoding is sufficiently performed in a turbo decoder, an excellent performance is obtained. However, in practice, the system obtains the performance required by the system by motivating that the code gain does not continue to increase even after further iterative decoding depending on the channel environment. In general, we considered Stop Criterion that controls repetitive decoding. Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings.

3 is a block diagram showing a configuration of an iterative decoding control apparatus of a turbo decoder according to the present invention, wherein a first decoder 304, an interleaver 306, a second decoder 308, a first deinterleaver 311, and a first decoder are shown. 2 includes the deinterleaver 313, the iteration decoding controller 317, the switching unit 320, the hard determiner 321.

After receiving the signal output from the turbo encoder of FIG. 1, that is, the information bit X _k 300 and the parity bit 301 through the channel, the first parity bit Y _k 301 is depunctulated. The parity bit y _1k 302 output from the encoder 101 is sent to the first decoder 304 and the parity bit y _2k 303 output from the second encoder 105 is sent to the second decoder 308. Thereafter, the first decoder 304 decodes using X _k 300 and y _1k 302, and then interleaves the output 305 of the first decoder 304 in the interleaver 306. The second decoder 308 decodes one signal 307 and y _2k (303).

At this time, if iterative decoding is not desired, deinterleaving the output signal 309 of the second decoder 308 by the second deinterleaver 313 and then hardly determining the signal through the hard decision unit 321. 322 is sent as a decode signal. However, in order to perform repeated decoding, the signal 312 obtained by deinterleaving the output signal 310 of the second decoder 308 by the first deinterleaver 311, the received signals X _k 300 and y _1k 302 Decode again to the first decoder 304 by using. This iterative decoding can be done until the desired performance is achieved.

However, since repeated decoding causes problems such as additional calculation amount and decoding delay as the number of repeated decoding increases, an additional decoding controller 317 is additionally configured to stop decoding after a predetermined repeated decoding to solve this problem.

FIG. 4 is a detailed view of the iterative decoding controller and the peripheral device of FIG. 3, wherein the first soft decision unit 401, the second soft decision unit 403, the counter 405, the first code storage unit 317-1, 2 Repeating decoding controller 317 and hard decision unit 321 including a code storage unit 317-2, a comparator 317-3, a down counter 317-4, and a repeat decoding stop signal generator 317-5. It includes.

The first soft determiner 401 soft-determines the output 318 of the first decoder 304 to output a sign bit of the LLR (Log Likelihood Ratio), which is a soft decision value, and the second soft determiner 403 performs a second decision. A soft decision on the output 315 of the interleaver 313 outputs the sign bit of the LLR, which is a soft decision value.

LLR means the logarithm of the ratio of the posterior probability (APP) when the input information bit is 1 (one) and the posterior probability value when the input information bit is 0 (zero), and is output from the turbo decoder. The definition for LLR is equal to Equation 1.

Estimated is the information bit d _k is 0 (zero) -, L ( d k) value is (+) if the information bit d _k is 1 (one) with, L (d _k) value by the equation (1) () The larger the LLR value, the higher the reliability of this estimation.

Therefore, the LLR defined in Equation 1 represents the reliability of the information bit d _k and plays an important role in improving performance during iterative decoding. Iterative decoding improves the LLR value of each information bit. Each information bit is determined by determining the sign of the LLR value. It is assumed that as the repeated decoding is repeated, the codes of the LLRs of the first decoder and the second decoder become as in Equation 2.

In Equation 2, i denotes the number of repeated decoding and k denotes the k-th input information in one frame. Also Is the LLR output of the first decoder 304, Denotes the LLR output of the second decoder 306. As the repeated decoding is repeated, the code bits of the LLRs of the first decoder 304 and the second decoder 305 converge to the same value, and the information bits can be decoded correctly.

The first code storage unit 317-1 stores the code bits of the LLR which are soft decision values for the output 318 of the first decoder 304 output from the first soft determiner 401, and stores the second code. The unit 317-2 stores the sign bit of the LLR, which is a soft decision value for the output 315 of the second deinterleaver 313 output from the second soft determiner 403.

However, since the LLR sequence of the first decoder 304 has an interleaver 306 between the second decoders 308 and thus has different output sequences, the first code storage unit 317-1 has a soft determined first decoder ( The output 318 of the 304 is stored and the second code storage unit 317-2 stores the output 315 of the second deinterleaver 313 that is softly determined.

The comparison unit 317-3 compares the LLR codes of the information bits stored in the first code storage unit 317-1 and the second code storage unit 317-2. As a result of the comparison by the comparison unit 317-3, if the LLR codes of the information bits stored in the first code storage unit 317-1 and the second code storage unit 317-2 are the same, 0 (zero) is output. However, as a result of the comparison of the comparison unit 317-3, if the LLR codes of the information bits stored in the first code storage unit 317-1 and the second code storage unit 317-2 are different, one (one) is output. .

The down counter 317-4 counts the value output from the comparator 317-3 in descending order. Here, the initial value of the down counter 317-4 is initialized according to the BER required by the system.

For example, when the initial value is set to "5" in the down counter 317-4, when the output of the comparator 317-3 is 1 (one), the down counter 317-4 counts the descending order. Since the count value becomes "4" afterwards, if the output of the comparator 317-3 is zero, the down counter 317-4 does not perform the descending order count.

When the value of the down counter 317-4 becomes 0 (zero) due to the descending order counting, the descending order count is not performed any more.

The repeat decoding stop signal generator 317-5 outputs a repeat decoding stop signal according to the values of the down counter 317-4 and the counter 405. If the value of the down counter 317-4 has a value other than 0 (zero) or the value of the counter 405 has performed the maximum number of times of repeated decoding, the iterative decoding stop signal generator 317-5 A repeat decoding stop signal is output to 321).

However, if the value of the down counter 317-4 has a value of 0 (zero) or if the value of the counter 405 does not perform the maximum number of times of repeated decoding, Control to perform repeated decoding continuously.

The iterative decoding stop signal generator 317-5 logically ORs the output of the down counter 317-4 and the output of the counter 405, and outputs the result to the hard determiner 321.

The repeated decoding stop signal 319 output from the repeated decoding stop signal generator 317-5 is also used as a switching control signal of the switching unit 320 in FIG. The output 314 of 313 is switched to the hard determiner 321.

The maximum number of repeated decoding max_N_iter is set in the counter 405, and the counter 405 counts the number of repeated decoding of the turbo decoder during the maximum number of repeated decoding and outputs it to the repeated decoding stop signal generator 317-5.

When the hard determiner 321 is turned on by the repetition decoding stop signal output from the repetition decoding stop signal generator 317-5, the second decision unit 321 may perform a second decoding operation for a predetermined number of repetition decoding times. The output of the deinterleaver 313 is hard-determined to output the decoded result. The hard determiner 321 outputs the hard decision value as 1 (one) when the value output from the second deinterleaver 313 is positive and 0 (zero) when the value is negative.

In the present invention, it is determined whether to repeat the decoding according to whether the LLR code bits of the first decoder and the second decoder of the turbo decoder converge. To determine convergence, the following four cases can be represented by the difference (L _sd ) of the code bits of the two decoders LLR and the number of errors in the frame (N _err ).

Case 1. L _sd and N _er have large values even if iterative decoding proceeds.

Case 2. L _sd and N _err become 0 as repeat decoding proceeds.

Case 3. L _sd is 0 as iterative decoding proceeds but N _err is not 0

Case 4. L _sd does not become 0 as iterative decoding proceeds, but N _err becomes 0.

First, case 1 is a case in which an error remains in the frame and the information bit cannot be decoded even though repetitive decoding is performed.

Case 2 is a case in which the information bits are correctly decoded because there are no errors in the frame as the repetition decoding proceeds. In this case, it is determined that the information bits will converge, and the repetitive decoding is stopped.

In case 3, the decoder judges that there is no error in the received information string, but an error remains in the frame. In this case, it is determined that the decoder will converge and stops repeated decoding.

Case 4 is a case where the LLR codes of the two decoders are not the same but there is no error in the frame. In this case, it can be regarded as a phenomenon that occurs because the second decoder corrects the error shown in the first decoder as the iterative decoding proceeds. Thus, in this case, if the repetition is stopped, it may be considered that the repetition is stopped just before Equation 2. On the basis of this, an interruption condition for stopping repeat decoding is given by Equation 3.

Where S _d is the number of signs of different LLR values of the first decoder and the second decoder, and is the frame size. The iterative decoding stop algorithm proposed in this paper uses different numbers of LLR codes of two decoding periods as a measure of the reliability of the LLR output of the turbo decoder during each iteration.

5 is a flowchart showing the operation of the iterative decoding control method of the turbo decoder according to the present invention. As shown in FIG. 5, the turbo code first sets the maximum number of repeated decoding (max_N_iter) of the counter 405 to control the repeated decoding, and initializes the current number of repeated decoding (N_iter = 0) of the counter 405. The initial value of the down counter 317-4 is set (step 501).

Subsequently, the first decoder 304 decodes the signal output from the turbo encoder (step 503), and the code bit of the LLR, which is a soft decision value for the output 318 of the first decoder 304, is first encoded. The data is stored in the storage unit 317-1 (step 505).

A soft decision value for the output 315 of the second deinterleaver 313 after decoding the signal output from the turbo encoder by the second decoder 308 (step 507), deinterleaving the second deinterleaver 313. The code bit of the LLR is stored in the second code storage unit 317-2 (step 509).

Thereafter, when iterative decoding is performed for performance improvement, the number of repeated decodings N_iter of the counter 405 that counts the number of repeated decodings is increased by one (one) (step 511).

The comparison unit 317-3 compares the LLR codes of the information bits stored in the first code storage unit 317-1 and the second code storage unit 317-2 (step 513).

However, if the comparison result of the comparison unit 317-3 (step 515) and the LLR code of the information bits stored in the first code storage unit 317-1 and the second code storage unit 317-2 are different, 1 (one ) Is output (step 517).

Thereafter, the down counter 317-4 decreases the set initial value by one (one) (step 519).

If the comparison result of the comparison unit 317-3 (step 515) and the LLR code of the information bits stored in the first code storage unit 317-1 and the second code storage unit 317-2 are the same, 0 (zero) is returned. Output (step 521).

In operation 523, the counter 405 determines that the down count execution result of the down counter 317-4 is zero (zero). It is determined whether the maximum number of iterations has been performed (step 525).

When the down count execution result of the down counter 317-4 is not zero (0) or the counter 405 performs the maximum repetitive decoding, a repeat decoding stop signal is generated (step 527), and the second interleaver ( Hard output is determined in step 313 and output (step 529).

However, if the counter 405 does not perform the maximum iterative decoding, it jumps to step 503 to perform the iterative decoding.

FIG. 6 is a table comparing average number of times of repeated decoding of a stop criterion by cross entropy according to the stop criterion of the present invention.

As a result of simulation of turbo code having 1/3 code rate and 8-state in the BER = 10 ^-6 region, the stopping criterion of the present invention is about 0.5 times to that of the conventional cross-entry method. It was confirmed that there was an effect of reducing the average number of repeated decoding about 1.1 times and reducing the amount of calculation about 11% to 19%.

So far I looked at the center of the preferred embodiment for the present invention. Those skilled in the art will appreciate that the present invention can be implemented in a modified form without departing from the essential features of the present invention. Therefore, the disclosed embodiments should be considered in descriptive sense only and not for purposes of limitation. The scope of the present invention is shown in the claims rather than the foregoing description, and all differences within the scope will be construed as being included in the present invention.

As described above, the present invention can reduce the calculation amount and the decoding delay time due to repetitive decoding, which is one of the problems of turbo code, and can be used to determine the repetition decoding interruption criteria. Is as follows.

First, according to the present invention, in the turbo decoding process, the LLR code for the soft decision output value of the first decoder and the second decoder is determined, and if the predetermined condition is satisfied, the iterative decoding is stopped, and even if the decoding result does not converge, There is an effect that the decoding delay of the data can be reduced by stopping after proceeding for the number of times of repeated decoding.

Second, according to the present invention, the power consumption can be reduced due to the shortening of the decoding delay time, the capacity of the entire system can be increased, and the high quality multimedia service can be provided.

Third, according to the present invention, by controlling the number of iterative decoding of the turbo decoding apparatus using the stop criterion of the present invention, the exponential required in the cross-entropy scheme of Hagenauer, which is a conventional iterative decoding technique, is calculated. Since this unnecessary and relatively simple calculation is performed, the amount of calculation can be reduced.

1 is a block diagram showing the configuration of a general turbo encoder.

2 is a block diagram showing the configuration of a conventional turbo decoder.

3 is a block diagram showing the configuration of an iterative decoding control apparatus of a turbo decoder according to the present invention.

4 is a detailed view of the iterative decoding controller and peripheral devices of FIG. 3.

5 is a flowchart showing the operation of the iterative decoding control method of the turbo decoder according to the present invention.

FIG. 6 is a table comparing average repeat decoding times of a stop criterion according to the present invention and a stop criterion by cross-entropy.

Claims (7)

As a turbo decoder, Soft decision value output means for iteratively decoding the turbo encoded data and outputting a soft decision value; Repetitive interruption of repeating the decoding of the turbo decoder by counting the case where the sign is different by comparing the codes of the soft decision values output from the soft decision value output means or when the number of repeated decoding exceeds the set maximum number of repeated decoding. Control means; And Repetitive decoding control apparatus of turbo decoder using LLR code comparison comprising hard decision value output means for hardly determining the output value of the soft decision value output means and outputting the decoded result according to whether repeat decoding of the repeat control means is performed. . 2. The counter according to claim 1, further comprising a counter for setting a maximum number of repeated decoding and counting the number of repeated decoding from the soft decision value of said soft decision value output means and outputting the result to said repeat stop control means. Iterative decoding control apparatus of a turbo decoder using the sign comparison of the LLR. The method of claim 2. The soft decision value output means A first soft decision unit for soft decision of the first decoded data; And And a second soft decision unit for soft decision of second decoded and interleaved data. The method of claim 1, wherein the repeat stop control means Comparing the codes of the soft decision values output from the soft decision value output means to count the case where the signs are the same, or if the number of repetition decoding times is less than the set maximum number of repetition decoding times, controlling to perform the repeated decoding of the turbo decoder. Iterative decoding control apparatus of turbo decoder using code comparison of LLR characterized in that. The method of claim 3, wherein the repeat stop control means A first code storage unit which stores a code of a bit string output from the first soft determiner; A second code storage unit which stores a code of a bit string output from the second soft determiner; A comparator for comparing codes of the first code storage unit and the second code storage unit; A down counter for counting down count values set in descending order when the codes of the first code storage unit and the second code storage unit are different as a result of the comparison; A counter for counting the number of times of repeated decoding; And LLR characterized in that it comprises a repeat decoding stop signal generator for outputting a repeat decoding stop signal to the hard decision output means when the down counter counts a non-zero value or the counter performs a set maximum number of repetitions Iterative decoding controller of turbo decoder using code comparison An iterative stop control method applied to a turbo decoder, (a) storing the sign of the soft-determined first decoded data; (b) storing the sign of the soft-determined second decoded and deinterleaved data; (c) performing down counting of the down counters by comparing the codes and counting the cases of different codes in descending order; (d) comparing a value of counting the number of times of repetition decoding by a counter with a maximum number of times of repetition decoding previously set in the counter; (e) generating a repeat decoding stop signal when the count value of the down counter is not 0 or the counter exceeds the maximum number of repeated decoding; And and (f) hard-determining and outputting the soft-determined second decoded and deinterleaved data. The method of claim 6, When the down counter counts the case where the signs are the same or when the counter counts less than the maximum number of repeated decoding times, the repeated decoding of the turbo decoder is continuously performed. Iterative decoding control method of turbo decoder.