Nanjing University of Posts and Telecommunications, Sun Li, Nanjing University of Posts and Telecommunications Zhang Zong Orange
Summary: Turbo code, due to performance close to Shannon theoretical limit, is better than other coded in the application environment of low signal-to-noise ratio. Therefore, in a variety of schemes of the third generation mobile communication system, it is considered that the TURBO code is one of the coding criteria of the wireless channel. This paper introduces the structure and codec method of Turbo code, and its application in the third generation mobile communication system. Key words: channel coding; Turbo code; RSC encoder; interleaver; iterative decoder; third-generation mobile communication Abstract: Turbo Codes, Because of it's outstanding Performance In Channel Coding, HAS BEEN CONSIDERED BY IMT-2000 AS A Plan Of . application This paper introduces the principle of basic structure of Turbo codes.Also, the paper introduces the mainly application in 3G (3ird Generation) mobile telecomminucation systems Key words: channel coding; Turbo codes; RSC; interleaver; recursive algorithm; 3G (. The 3ird generation) Mobile Telecommunication Systems 1. Introduction Information Theory has been born for more than 50 years, people have been trying to find a coding method that is more close to Shannon, error probability. At the 1993 ICC International Conference, C. Berrou, A.Glavieux and P.thitimajshiwa proposed a comparative, decoding scheme called Turbo Code, and in the interleaver size: in the case, iteration 18 times, It has computer simulation. The simulation results show that when it is alive, the signal to noise ratio is changed. Its coding gain is less than 1 dB than the gap between the Shannon channel capacity. [1] From the third generation of mobile communication system candidates, it is generally required to provide medium-speed or high-speed data services, and the general data traffic channel is 64kbps, 144kbps, 384kbps. In the radio channel with low signal-to-noise ratio, the Turbo code is more preferably more preferably, comparing cascaded RS plus volume codes used in the second generation mobile communication, Turbo code performance can increase 1 dB. So in the IMT-2000 scheme, the Turbo code has been used as one of the channel coding standards for transmitting high-speed data. [3] [4] This article introduces the basic structure and compiling coded principle of Turbo code, and the application of Turbo code in third generation mobile communication. Second, the title of the Turbo code, the coding principle of the decoding principle It can be seen from the encoder schematic. The code is composed of three parts: two RSC subcoders RSC1, RSC2, an INTERLEAVING, one PUNCTURING unit. 2.1 The encoder design assumes that the information unit of the input information encoder is, which directly inputs RSC1 directly to encode, generates the check sequence X1p. On the other hand, after interleaving, an interleaving system sequence u 'and another check sequence X2p are produced.
When the code rate R is greater than 2/3, in any size signal-to-noise ratio, since the free distance of the system recursive convolutional encoder RSC (Recursive SystematicConvolutional) is larger than the non-system volume, BER is BER. Small, showing better performance, and the RSC encoder [1] is used in the Turbo code [1]. Its encoding matrix can be represented: where it is: The encoder consists of a memory cell that is input to an information bit. To make the encoder initial state in the full state, you need to increase the M bit information (not necessarily zero) after the information sequence, and the two encoders are simultaneously set, or the two ways not intertwined. Delay. 2.2 Design interleaver of the interleaver is typically read backwards after randomly replacing the input original information sequence. The role of the interleaver is: First, the long code can be generated. Second, the input of the two RSC encoders is not related, the encoding process tends to stand independent. Interleaving makes the encoding to generate an endlessness, make the code random, homogenize, and directly affect the performance of the Turbo code directly. In the decoder, for a sub-decoder, it is not corrected for an unconceder, and after the other decoder is dispersed, it is a corrected error. The interleaving mode is mainly ruled to interleave, irregular interleaving and random interleaving. Usually the rule is interleaving, the process is written, and the effect is not good. Random interleaving fingering format is randomly assigned, the best interleaving mode in theoretical performance, but since the entire interleaving information location information is to be transmitted to the decoder, the coding efficiency is reduced. In practical applications, irregular interleaving is generally used, which is a pseudo-random interleaving method that uses a fixed interleaving method for each encoded block, but the interleaver structure between blocks and blocks is different. It is often necessary to obtain the length of the interleaver for the high coding gain. The wireless mobile communication system is high in time delay, so a pseudo-random short interleaver with an interleaving length of 400 is used. 2.3 Practice Unit If we first design a low yard yard, some of the check bits (ie, fracture) deleted during transmission, let it become a high-code rate code, and avoid high-code rate rolls through this way. Calculation complexity inherent in the coding decoding operation. The cracking process can be described as a selected bit periodically from the encoder output, so that a mesh code that periodically changed. For example, the sub-encoder code rate is, by deleting all of the oddle and the bits, the yield of the entire encoder becomes r = 1/2. When the code rates of the subcoders C1, C2 are different, the entire encoder code rate R and the self-encoder C1, C2 code rates R1, R2 are: -1 Turbo code iterative decoding Figure 1-2 gives A Turbo Code decoder of a feedback structure, due to the existence of interleaving links, it is impossible to have a truly feedback, but the water line iterative structure. It is also this pipeline structure such that the decoder can be constructed from a number of exactly the same software into the base unit. The decoder working principle is: the received serial data is converted string, while filling the fill the fill bit (ie, does not affect the value of the decoded decoding, such as 0). The check sequence R1 generated by the information sequence R0 and the RSC1 is sent to the soft output decoder 1, and the external information sequence Z1K generated by the soft output decoder 1 is intertwined as the input of the next soft output decoder 2. The information sequence r0 is input to the decoder 2 through the interleaver, and the check sequence R2 generated by RSC2 is input simultaneously. The output information Z2K of the decoder 2 is used as a feedback to the decoder 1 after the deinterleaver, and the above process is repeated, until the final decoding output is no longer improved, and the final result is decoder 2 The output is deinterleaved as a decision output.
The advantage of this decoder structure is that each decoder can not only utilize the information bits and check bits of this Code, but also use the information provided by the previous decoder to decode, thereby improving the decoding accuracy. Sex. Its disadvantage is that iterations take more time, resulting in the TURBO code to be restricted in some communication systems (such as digital phones, etc.) that are highly delayed. The TURBO code decoding algorithm is based on the maximum post-test probability (MAP) algorithm or is soft and exported to the Viteri (SOVA) algorithm. The MAP algorithm is minimizing symbol or ratio error probability, and the SOVA algorithm is minimizing sequence error probability. In a low SNR environment, the MAP algorithm has improved than the performance of the SOVA algorithm, but the MAP algorithm considers all paths at every moment, and its operation is multiplication and index operation, complicated. The operation in the SOVA algorithm is a simple addition operation, comparison and selection. TI added a comparison selection structure suitable for the Viterbi algorithm in its DSP chip C54, C55, so that the SOVA algorithm is easier to implement hardware. Third, the application channel coding technology in the third generation wireless communication can improve the system reliability during the transmission process due to noise and interference during transmission. Therefore, it is one of the key technologies in the 3G mobile communication system. The diversity of the business types provided by the 3G mobile communication system, which puts higher requirements for error control compiling codes. The WCDMA and CDMA2000 programs are recommended to use Turbo encoding techniques in addition to the convolutional coding techniques and interleaving techniques similar to the IS-95 CDMA system. [3] [4] 3.1 In the design of the RSC encoder, the TURBO code is used in front of the CDMA system and in the reverse link channel. In the reverse link channel, the generated matrix of the subcoder (3, 1, 3) RSC is: g (d) = RSC encoder based on 8 state-based parallel coil catalysis (8PCCC). Interleaving adopts bit flip technology. The TURBO code of 1/4, 1/2, 1/3 is adopted by the closing treatment. [3] The output parity v2 and v2 'of the two sub-encoders alternately deleted, and the TURBO code having a code rate is 1/4; for V1, V1', the code rate is 1/3; For V2, V2 'interval V1, V1', can be deleted, and the code rate is 1/2. In the forward link channel, the generation matrix of (3, 1, 3) the RSC encoder is: g (d) = alternately deleted the output parity V1 and V2 'of the two sub-encoders, and can obtain the code rate 1/4 Turbo code; for V2, V2 ', the code rate is 1/3; 1/2 rate can be obtained as the reverse link channel. In WCDMA, the RSC sub-encoder uses 8-PCCC using 8-state parallel coin volumes 8-PCCC for data services between business service quality requirements. [4] The generated matrix is: g (d) = 3.2 Selection of interleaving lengths The decoding performance of the Turbo code is largely dependent on the intersectment depth. In 3G mobile communications, the business rate is 32kbit / s to 2 mbit / s. 10MS one frame, the frame length is from 20 to 20,000. In order to improve decoder performance, in some low speed services, multiple frames can be used to form a data block to increase the intersecting depth. In WCDMA, the Turbo codeware is a septic sequenal block interleaver. The number of interleaving lines is 5, 10 or 20 lines, and the number of columns is selected on the basis of the number of lines.
The data is read into the interleaver, press the fixed mode to perform line conversion, and the length of different input sequences corresponds to different rows and line conversion modes. After the conversion is completed, column conversion is performed. Different lines correspond to different columns of conversion parameters, take close to randomized prime taking algorithms. After the data is completed, it is read by column. [4] CDMA2000 is also based on block interleaving. The number of interleaving lines is 25 = 32 rows, and the number N = 2N, N is satisfying the minimum of 32N greater than or equal to the length of the frame length. Data is read in line. The basis of line conversion is the principle of Bit flipping. The inter-column conversion replacement formula is: x (i 1) = [x (i) c] mod n is the same as an empty method, in order to closer to randomization, make each column of offset different values. After the data is converted, the data is output. 3.3 Decoder Design Due to the complexity of the Turbo code method, the decoding delay is long, so the data service application with high delay is limited. Therefore, the low complexity decoder is designed to become the focus of the Turbo code decoding algorithm. In order to exchange a simplification of complexity, the presence of secondary performance is allowed to decode. For example, the decoding of the Turbo code is allowed in 3GPP than the standard MAP algorithm has 1 dB gain loss. Combined with CRC checks to reduce the number of iterations, reduce decoding complexity and decoding delay when SNR is large. [5] 3.4 Turbo code and other techniques combined with Turbo code developed, Turbo code and Turbo ideas are increasingly used in combination with other technologies in the 3G mobile communication system design. For example, in CDMA, since the use of interleaving is used in Turbo code encoding, the correlation between the spreading code can be reduced by the position of the dispersion information symbol. When implemented, the Turbo code is combined with the spread spectrum coding of the DS-CDMA system, the spreading code is the internal code, the Turbo code is external code, connected to the interleaver, similar to the form of a cascading code. The receiving end, first separate the received information of each user, then calculate the passage probability of the transfer condition according to the channel model, after the branch probability generator obtains the subkey probability of receiving information, respectively, respectively, respectively The Turbo code decoder, each decoder gets a soft decision output and a side information, where the branch probability generator is sent back to the branch probability generator, which is an iterative translation of the next iterative decoding. code. In addition, Turbo code and modulation combined T-TCM, Turbo code and ARQ, Turbo code iterative ideas were used for multi-user detection, and the like technology became a hot spot in 3G mobile communication technology research. Fourth, the study of Turbo code is still lacking in theoretical infrastructure, but in various harsh conditions (i.e., low SNR), communication capabilities that provide access to the Shannon limit have passed analog proof. Commercial 3G mobile communication system is still in the research and development phase. As one of the key technologies, the Turbo code will gradually obtain better theoretical support and is further developed and improved.
