KR100407328B1 - Channel coder of mobile communication system and encoding method thereof - Google Patents

Abstract

PURPOSE: A channel coder of a mobile communication system and an encoding method thereof are provided to transmit frame data of a low transmission rate without preform deterioration by using a repeater. CONSTITUTION: A channel coder of a mobile communication system includes a repeater, a first coder(115), a circular shift interleaver(113), and a second coder(117). The repeater generates an input sequence having symbols corresponding to a frame standard of N information bits by repeating input frames as much as a predetermined number. The first coder(115) is formed with an RSC coder in order to encode the output of the repeater and generate the first parity. The circular shift interleaver(113) is used for changing orders of information bits of the repeater. The second coder(117) is formed with the RSC coder to encode the output of the circular shift interleaver and generate the second parity.

Description

Channel Encoding Device and Method in Mobile Communication System

The present invention relates to a coding apparatus and method in a mobile communication system, and more particularly, to a channel coding apparatus and method using an iterator and a turbo coder.

Currently, a coding scheme used in a signal processor of a digital mobile communication terminal uses convolutional code, and in order to maintain a constant data rate, a frame standard of data having different data rates is a frame of voice data having a set data rate. Frame data is repeatedly transmitted in order to be identical to the standard.

In general, it is known that the coding gain increases as the length of the codeword or the memory of the encoder increases. However, this also increases the complexity of the maximum likehood decoding algorithm and makes it difficult to implement in hardware. In particular, the Viterbi algorithm, which is used as a decoding method for convolutional codes, has an exponential complexity with an increase in the constraint length K, making it difficult and impractical for K> 8. Therefore, new coding schemes and decoding schemes are needed to increase the performance of convolutional code with short constraints.

Turbo codes are described in 1993 by Berrou [C. Berrou, A. Glavieux and P. Thiitimajshima, "Near shannon limit error-correcting coding and decoding: turbo codes," Proceedings of ICC `93, pp.1064-1070, Since its publication by Geneva, Switzerland, May 1993.], its research has been focused on breakthrough performance. Turbo code performance using random interleavers with sizes of tens of thousands is known to approach the Shannon limit. However, in order to use a turbo code using a very large random interleaver, a very high bit rate and a long delay must be taken. Therefore, the turbo code has a problem that is difficult to apply to other applications except for deep space communication.

In addition, the turbo code using the symbol repetition and interleaving after the interleaver is encoded and encoded therein has a problem that performance is remarkably degraded when the frame size is 10 ms or less (50 bits or less per frame).

Accordingly, an object of the present invention is to provide an encoding apparatus and method capable of transmitting data at a low data rate and frame size in a voice data transmission apparatus of a mobile communication system.

Another object of the present invention is to add a repeater before the turbo encoder in the voice data transmission apparatus of the mobile communication system to repeat the symbol of the frame data having a low data rate and apply it to the encoder to transmit the voice without a performance reduction An apparatus and method are provided.

It is still another object of the present invention to add a repeater after a turbo encoder in a voice data transmission apparatus of a mobile communication system to encode frame data having a low data rate, and then repeat the encoded data to meet a frame rate of a set data rate. The present invention provides an apparatus and method for transmitting the same.

In order to achieve the above object, a channel encoding apparatus of a digital mobile communication system according to an embodiment of the present invention repeats a bit in a set number of times for an input frame to generate an input sequence having symbols of frame information of N information bits. And an RS encoder, comprising: a first encoder for encoding the output of the repeater to generate a first parity; and a bit order of information output from the repeater, wherein Addr [i] = (p * i) an interleaver interleaving with mod N (N: frame size, p is a natural number that satisfies gcd (N, p)), and an RS encoder, which encodes the output of the interleaver to generate a second parity The encoder and a switch for switching the output of the first encoder and the second encoder at a half rate.

1 is a diagram showing the configuration of a channel encoding apparatus according to a first embodiment of the present invention.

FIG. 2 is a diagram showing the configuration of the turbo coder of FIG.

3 is a diagram illustrating a configuration of an apparatus for decoding an output of a turbo coder having the configuration as illustrated in FIG. 2.

4 shows error rate performance test results for AWGN channels.

5 is a diagram illustrating error rate performance test results according to frame positions.

6 shows error rate performance test results for a Rayleigj fading channel.

7 is a diagram showing the configuration of a channel encoding apparatus according to a second embodiment of the present invention.

FPLMTS can provide a variety of services. Therefore, the FPLMTS terminal should be designed to satisfy the transmission rate, maximum BER, and maximum delay required for each service. Looking at the service to be provided by the FPLMTS, can be largely divided into voice service and data service. The voice service uses low data rates such as 8 kbps and 32 kbps, has a delay of less than 40 ms, and requires BER &lt; 10 ^-4 . The data service uses high data rates such as 144 kbps, 384 kbps and 2048 kbps, with a delay of around 200 ms and requiring BER <10 ^-6 .

In a frame-based system, the delay is directly related to the size of the frame. Thus, re-expressing the requirements of FPLMTS, the FPLMTS system allows high BER for small frames and low BER for large frames. Turbo codes can achieve increasingly lower BER by increasing the size of the interleaver used. Therefore, the turbo code is the best channel code for the IMT-2000 because it can satisfy the requirements of the IMT-2000 by changing only the size of the interleaver without changing other parts.

For example, the turbo code can achieve lower BER by increasing the number of iterative decoding. If the user wants a better quality call than the service provider sets, the user only needs to increase the number of repeated decoding of his terminal. Conversely, even if the call quality deteriorates, if you want to use the battery for a long time, you can reduce the number of repeated decoding.

1 is a diagram showing the configuration of a channel encoder according to an embodiment of the present invention. Referring to the configuration of FIG. 1, the MUX10 multiplexes input information data and other data and outputs the multiplexed information. CRC20 adds a CRC to the data output from the MUX10. The symbol repetition 30 inputs frame data, repeats a bit a predetermined number of times, and outputs each symbol to generate an input sequence U. The input sequence U is transmitted on a channel. The turbo encoder 40 encodes and outputs the data output from the repeater 30 according to a transmission standard. The turbo code encoder 40 generates a parity symbol using an input consisting of a frame of N information bits. Channel interleavers 50 and 60 channel interleave the data encoded by the turbo code encoder 40. The MUX60 and 80 output data interleaved in the corresponding channel interleavers 50 and 60 on the channel.

In the channel encoding apparatus of the mobile communication system according to the first embodiment of the present invention having the structure as shown in FIG. 1, the turbo code encoder 40 has the structure as shown in FIG. Referring to the configuration of FIG. 2, the first encoder 115 is a RSC coder (Recursive Systematic Convolution coder), and outputs the input sequence U output from the repeater 30 as the output sequence y _1k of the RSC. The interleaver 113 inputs the input sequence U output from the iterator 113 and distributes the interleaver 113 so as to remove the correlation of aggregation errors. The second encoder 117 is a RSC coder (Recursive Systematic Convolution coder), and outputs the sequence output from the interleaver 113 as the output sequence y _2k of the RSC. The switch 119 alternately outputs the outputs of the first encoder 115 and the second encoder 117 according to the set transmission rate.

Referring to FIG. 2, a repeater 111 repeatedly transmits input frame data to a turbo encoder. When the repeater 30 transmits a voice having a frame size of 5 ms and a data rate of 8 kbps, the 5 ms frame is repeated one bit twice in succession and then outputs each symbol to the turbo code encoder 40.

The turbo code encoder 40 includes an interleaver 113, a first encoder 115, a second encoder 117, and a switch. The first encoder 115 and the second encoder 117 of the turbo code encoder 40 have a structure in which two simple recursive convolutional codes that make a parity symbol using an input composed of a frame of N information bits are connected in parallel. . The first encoder 115 inputs the output of the repeater 11. The interleaver 113 inputs the output of the repeater 30 and changes the order of the information bits input to the second encoder 117 and has the same size as the frame of the N information bits. Accordingly, the output of the turbo code encoder 40 has dual parity information due to the output of the cyclic convolutional encoders 115 and 117 as well as the output modified through the interleaver 113.

Referring to the operation of the turbo code encoder 40, when the input sequence U output from the repeater 30 is represented by Equation 1, each cyclic systematic convolutional code (RSC) output from the first encoder 115 and the second encoder 117 : The output sequence of Recursive Systematic Convolutional Code) is shown in Equations 2 and 3 below.

[Equation 1]

[Equation 2]

[Equation 3]

In the turbo code encoder 40 structure having the configuration as shown in FIG. 3, GO ₁ , G ₁ , GO ₂ , and G ₂ each represent a generation polynomial of the configuration code. The y _1k and y _2k , which are outputs of the encoders 115 and 117, are alternately transmitted.

As described above, the turbo code encoder 40 has a structure in which the first encoder 115 and the second encoder 117, which are recursive systematic convolutional encoders, are connected in parallel. In this case, when each bit is symbol-repeated through the repeater 30, the input sequence U may be expressed as Equation 1 above. The input sequence U is separated into a part transmitted to the channel and a part transmitted to the first encoder 115 and the interleaver 113. At this time, the interleaver 113 distributes the input sequence U to remove correlation of the aggregation error and then applies it to the second encoder 117.

The reason why the interleaver 113 is used in the turbo code encoder 40 is that when a burst error occurs, the output of the MAP decoder has a correlation, and in the second MAP decoder of the next step, the correct decoding is caused by the correlated input. Will not be able to. Therefore, it is essential to use an interleaver that can disperse the concatenation error in one frame so that it does not matter. In the embodiment of the present invention, it is assumed that the interleaver 113 uses a cyclic shift interleaver.

3 shows a decoder configuration of a turbo code. The turbo decoder has a structure for iterative decoding using the first MAP decoder 217 and the second MAP decoder 223. During repeated decoding, the additional information z _k is exchanged between the MAP decoders 217 and 223. The additional information initially has a value of 0, but as decoding is repeated, the additional information becomes more accurate, thereby improving error correction.

Since the MAP algorithm that outputs the Log-likelihood Ratio (LLR) requires a lot of computation and memory, the E function shown in Equation 4 is used to simplify the computational speed and memory usage.

[Equation 4]

When the LLR is represented by using the E function as shown in Equation 4, Equation 5 may be expressed as follows.

[Equation 5]

Here, the state metrics A ⁱ _k (m) and B ⁱ _k (m) can be written as Equation 6 below.

[Equation 6]

In Equation 6, S ^j _b (m) means a state where the next state is m when the input is j, and S ⁱ _f (m) indicates the next state when the input is i and the current state is m. it means. The branch matrix D _j (R _k , m) for the AWGN channel is expressed by Equation 7 below.

[Equation 7]

In Equation (7), σ ² is noise variance, x _k is an information signal received at time k, y _k is a parity signal received at time k, and Y ^j (m) is an input j and state When m is the output of the encoder.

By introducing the E function as described above, most operations of the MAP algorithm are converted into additions and calls to the E function. Since f (z) = ln (1 + e ^-z ) converges rapidly to zero, the E function can be implemented with a look-up table or a simple circuit. That is, in the 8-bit MAP decoder, only eight of the values of f (z) have a non-zero value. In the 4-bit MAP decoder, all values of f (z) are zero. Therefore, the MAP decoders 217 and 223 can be implemented very simply.

An operation in the case of using the turbo code encoder as the channel encoder of the mobile communication system will be described. Here, an example in which the turbo code encoder is applied to the IMT-2000 will be described.

First, the IMT-2000, the IMT-2000 has been studied for the purpose of commercialization in the early 21st century as a third generation mobile communication. At present, the specification of the IMT-2000 is not determined and only the requirements are published. Looking at the requirements, the IMT-2000 should provide a variety of services including multimedia data as well as voice. Table 1 below shows transmission requirements of services that are to be provided by IMT-2000.

TABLE 1

It is not appropriate to use the same channel code for services with different rates, acceptable delays, and error rates. In order to provide various services using one channel code, the channel code should be designed according to the most severe conditions. Therefore, IMT-2000, which is supposed to provide various services, should use adaptive channel code that can operate differently for each service.

If the service that is supposed to be provided in IMT-2000 is largely classified, a voice service with a low error rate and a required error rate of 10 ^-4 instead of a low allowable delay and a high allowable delay and a long allowable delay are required. It can be divided into a data service with an error rate of 10 ^-6 . In other words, it can be represented as a service using a short frame requiring an error rate of 10 ^-4 and a service using a long frame requiring an error rate of 10 ^-6 . This is in good agreement with the characteristics of the turbo code that the larger the frame size, the lower the error rate.

An embodiment of the present invention looks at the process of using a turbo code as the channel encoder for FPLMPTS. As described above, the turbo code encoder includes an interleaver 113 having the same size as the frame size. Therefore, before designing the turbo code for the FPLMPTS, it is necessary to determine the frame size. Here, only the specification for 8kbps voice service is presented. It is assumed that the 8kbps voice service frame specification shown in Table 1 has a depth of 10ms, and the actual transmission rate is 9.6kbps.

TABLE 2

In Table 2, the information data (information data) is 80 bits, the entire voice service frame is 96 bits. The CRC is inserted for error checking for ARQ when transmitting data requiring BER < 10 ^-6 in a voice band. In fact, the size of the interleaver 113 used for the turbo code is the value except the tail bit and the extension 0s.

Among the requirements of the IMT-2000, there is a protocol for a transport channel. Therefore, the FLMPTS coding system must satisfy the required BER and allowable delay in the transport channel required by IMT-2000. In the embodiment of the present invention, it is assumed that an equalizer exists in a receiver, and an object of the present invention is to design an encoder that satisfies a transmission condition required by IMT-2000 in an AWGN channel and a Rayleigh fading channel. To this end, a description will be given of a code system that can achieve the same performance as a half rate K = 9 convolutional code used for mobile communication.

As the performance of the turbo code is an interleaver, it is known that the best performance can be obtained by using a structure of a random interleaver. However, when the frame size is small, there is little chance that the randomly selected interleaver will perform well. Therefore, when the size of the frame is small, it is advantageous to use a structured interleaver, which is guaranteed to be more than the minimum weight, rather than using a random interleaver.

Structured interleaver can be classified into two-dimensional interleaver and one-dimensional interleaver. The 2D interleaver interleaves the frames as M * N arrays. Here, a block interleaver and a helical interleaver are included. Since the performance of the 2D interleaver is determined by M and N, a frame size is limited in order to use the 2D interleaver having excellent performance.

Representative one-dimensional interleavers include S.Dolinar and D. Divalar, "Weight distributions for turbo codes using random and nonrandom permutation," TDA progress Report 42-122, pp.56-65, August 15, 1995. have. The cyclic shift is interleaved in the same manner as in <Equation 8>.

[Equation 8]

In Equation 8, N is a frame size, and P is a natural number satisfying gcd (N, p) = 1. The cyclic shift is determined by p rather than size N of the interleaver.

In the embodiment of the present invention, the interleaver 113 uses a cyclic interleaver for use in voice service. Since the value of the extension 0s can be 0 and the tail bit can be 4, the size of the interleaver 113 is 92. p examines the information bits with a minimum weight of 4 or less to select the value with the largest minimum weight. In order to be able to decode the turbo code with one MAP decoder, a turbo code with G ₁ = GO ₁ = 7 and GO ₂ = 5 is used. The number of repetition numbers is 16.

Figure 4 shows the results of the BER performance test for the AWGN channel. Referring to FIG. 4, when the MAP decoder is processed with 8 bits, there is almost no difference in performance from when the actual value is used. However, when processing with 4 bits, the complexity is more than doubled than when processing with 8 bits, but there is about 0.3dB of loss compared to using the actual value. Therefore, the decoder of the turbo code for IMT-2000 is preferably 8 bits. Comparing the complexity of the 8-bit MAP decoder and the Viterbi decoder with K = 9 using 3 bits, the complexity of the MAP decoder is much smaller.

5 shows the results of the BER performance experiment according to the frame position. Referring to FIG. 5, it can be seen that the error rates at the beginning and the end of the frame are smaller than the average error rate. And even in the middle of the frame, the error rate for any position is less than the average error rate. This is because the coded sequence of the MAP algorithm decodes using the fact that state 0 ends in state 0. The relatively low error rate in the middle of the frame is where the frame starts and ends after interleaving. The reason why the first bit of the frame has a very low error rate compared to other parts is that since the interleaving method is a cyclic shift method, the first bit is still the first bit after interleaving. Therefore, where the error rate is low, such as at the beginning of a frame, it is better to transmit very important data, such as control bits.

6 shows the results of BER performance experiments for a Rayleigh fading channel. Referring to FIG. 6, although the turbo code encoder according to the embodiment of the present invention has similar performance to the convolutional code of K = 9 in the AWGN channel, there is an improvement of about 0.6 dB in the Rayleigh fading channel. Able to know. This is because the interleaver 113 is included in the turbo code encoder 40, so that good performance can be obtained for short fading. In the case of a 4-bit MAP decoder, BER performance is poor when Eb / No is small. This is because large noise is added, and most of the overflow occurs during the calculation, making it impossible to make accurate calculations. However, BER <10 ^-3 shows similar performance to K = 9 convolutional code and about 0.5dB loss for 8-bit MAP decoder.

7 shows a configuration of a channel encoding apparatus according to a second embodiment of the present invention. FIG. 7 has a configuration in which the repeater 30 repeatedly outputs the parity bits after generating the parity bits in the turbo code encoder 40.

The digital mobile communication system can provide various services. Therefore, the channel code should be designed to satisfy the transmission rate, delay condition and BER condition required for each service. In this case, a turbo code capable of satisfying service requirements by changing only the size of an interleaver may be most suitable as a channel code for use in a digital mobile communication system. The BER can be changed by adjusting the number of repeated decoding.

In an embodiment of the present invention, a turbo code for an 8 kbps voice service is presented. Comparing the turbo code with the convolutional code of K = 9 used in mobile communication, the decoder has much smaller complexity and shows similar performance for the AWGN channel and a gain of 0.6 dB for the Rayleigh fading channel. . As described above, since the interleaver is included in the turbo code encoder, good performance can be obtained for short fading.

Also, in the embodiment of the present invention, only the channel code for the voice service of 8 kbps has been described. However, when the frame standard is set and the interleaver is designed by the method according to the embodiment of the present invention for other services of the mobile communication system, Can be designed.

Claims (8)

In the channel encoding apparatus of the digital mobile communication system, A repeater for repeating the input frame a predetermined number of times and generating an input sequence having symbols of frame information of N information bits; An RS encoder, comprising: a first encoder for encoding a output of the repeater to generate a first parity; A cyclic shift interleaver for changing a bit order of information output from the repeater; An RS encoder, comprising: a second encoder for encoding the output of the interleaver to generate a second parity. The method of claim 1, wherein the cyclic shift interleaver, And addr [i] = (p * i) mod N (N: frame size, p is a natural number satisfying gcd (N, p)). The method of claim 1, wherein the channel encoding apparatus, And a switch for outputting the outputs of the first encoder and the second encoder at a half rate. In the channel encoding apparatus of the digital mobile communication system, An RS encoder, comprising: a first encoder for encoding input frame data to generate a first parity; A cyclic shift interleaver for changing a bit order of information output from the repeater; An RS encoder comprising: a second encoder for encoding the output of the interleaver to generate a second parity; And a repeater for repeatedly outputting the input frame data and the parity data output from the switch a predetermined number of times. In the channel encoding method of the digital mobile communication system, Generating repeating data symbols having a frame specification of N information bits by repeating bits for a set number of times for an input frame; Interleaving the bit order of the repetitive data symbols in a cyclic shift manner; And generating a first parity by RSS encoding the repetitive data symbols and generating a second parity by RSS encoding the interleaved symbols. The method of claim 5, wherein the cyclic shift interleaver method, And addr [i] = (p * i) mod N (N: frame size, p is a natural number satisfying gcd (N, p)). The method of claim 1, wherein the channel encoding method comprises: And outputting the first parity and the second parity at a half rate and combining the data symbols with the data symbols. In the channel encoding method of the digital mobile communication system, Interleaving the bit order of the input frame symbols in a cyclic shift method; Generating first parity by performing RS encoding on the input frame symbols by repeating the repeated symbols, and generating second parity by performing RSS encoding on the interleaved symbols; And repeating the bits as many times as the set number of times of the input frame symbols and the output parity symbols, and generating symbols of a frame standard of N information bits.

Images