Error correction coding type digital transmission method

Description: TECHNICAL FIELD [Detailed description of the invention]

【0001】

[Technical field to which the invention belongs]

The present invention relates to an error correction coded digital transmission method, and more particularly to a digital transmission method for a digital transmission system on a channel with significant perturbations. Furthermore, the present invention relates to an improvement of an error correction coded digital transmission method using a turbocode type coding method, which enables estimation of the quality of decoded information.

【0002】

[Conventional technology]

Digital transmission systems carry information using physical media such as propagation over cables, fiber optics, or wired channels. Such physical media are represented by the term "channel". In general, such a system has a channel coding device for suppressing channel defects, especially at the transmission level, and a decoding device at the destination level.

【0003】

The channel coding device is an error correction coding device. The error correction coding function consists of generating redundant information at the time of transmission, and this redundant information is damaged by the perturbations that occur during decoding at the destination, especially on noise, attenuation, and coherent channels. It is possible to reconstruct useful transmission information from the received information that reaches the destination, which is called reception information. A digital transmission method using such channel coding related to the corresponding destination decoding is called an error correction coded transmission method.

【0004】

For example, the coding operation is performed at the bit level. This behavior associates the binary source information sequence with the corresponding binary coded information sequence. When the size of the binary source information sequence is fixed, this binary coded information sequence is called a codeword. The binary coded information sequence is larger in size than the binary source information sequence so that redundancy can be introduced. Due to this redundancy, only specific coding information sequences that conform to the coding method are possible. If the received information sequence to be decoded is different from these possible sequences, this is because the received information sequence corresponds to the information damaged by the channel. Therefore, the task of the decoding method is to know the coding method used and to reconstruct the useful transmission information from the received information sequence as well as possible (the limiting factor is generally the complexity of the decoder). is there. Ideally, for optimal decoding in the sense that it is most likely, the sequence supplied by the decoder is the most promising sequence of all possible sequences.

【0005】

As the set of coding and decoding actions improves the ability to discriminate between sequences, so does the ability to correct errors.

【0006】

The performance of error-corrected coded transmission is generally a predetermined signal-to-noise ratio E._{b b}/ N₀(E_{b b}Is the energy per bit of information, N₀Is measured in bits or packet error rate relative to (is the power spectral density of noise). The code is given by using this_{b b}/ N₀It is described more effectively or less effectively, depending on whether the error rate is lower or less for the ratio and the complexity of the given decoding.

【0007】

Code efficiency is the number of useful information bits relative to the number of encoded information bits. In general, the lower the efficiency, the stronger the sign.

【0008】

Among the known error correction codes is a block code. Block coding consists of associating each block of k information bits with n-bit (n> k) blocks containing (nk) redundant bits. A block of n bits is obtained by multiplying a block of k source bits by a matrix of k rows and n columns called the code generator matrix. By permutation, the generator matrix is written in a format that indicates the identification matrix, so in the n-bit block, the k information bit and the nk redundant bit are separate, and the code is called a systematic code. The efficiency of the code is equal to k / n. The decryption device detects the error and corrects it. Such error detection codes well known in the art are, for example, Hamming code, BCH code, and Reed-Solomon code.

【0009】

It is also well known that error correction coding is performed by one or more convolutional coder. These operating principles are such that a device with a shift register type register encodes a block of k2 elements present at the time of input of a coder into a block of nbinary elements, taking into account the m block before the block existing at the time of input. It consists of becoming. The output of a convoluted coder consists of an n-coded binary element produced by the product of the convolution of the k2 element present at the time of input and the coder's response defined by the n generator polynomial. The number n of the coder's generator polynomials is called the coder's dimension. The efficiency of the code is equal to k / n. Decryption devices can be, for example, by continuous decoding, decoding by the most promising symbols, or the most promising sequence, as described in the reference "Digital Communications" by JG Proakis, published by MacGraw-Hill in 1995. Reconstruct the original data by decryption. For example, the Viterbi algorithm provides optimal decoding with the most promising sequences.

【0010】

According to this type of code variant, the encoding is stored in a shift-registered device rather than directly considering the set of m-source information items that precede the encoded information, each of which is available at input. This is done by using a series of m-auxiliary information items obtained by a mathematical combination of useful information items and a pre-computed m-auxiliary information item. Such a convolution code is called a recursive code. Furthermore, when useful information appears as being located in the n output of the coder, the resulting code is called the recursive systematic code represented by RSC (Recursive Systematic Convolutional).

【0011】

It is also known to associate different coder with each other in order to improve the coding performance. For example, the data encoded by the first coder can feed the second coder. Decryption is done symmetrically, starting with the second code.

【0012】

One effective type of coder association is, in particular, the article "Near Shannon Limit Error-Correcting Coding and" by C. Berrou, A. Glavieux, P. Thitimajshima, published in ICC-1993. Decoding: Turbo-codes , meeting minutes, as described on pages 1064-1070 are proposed. This type of coder association created a family of coding schemes known in the art called turbocodes. Turbo codes are based on an association of many simple codes, called basic codes, called concatenation, with the intervention of permutation actions called interleaving, which changes the order in which data is considered by each of the simple codes. The error correction code is shown. For example, interleaving can be obtained in the traditional fashion by an interleaving matrix in which data is inserted row by row and restored column by column. Such interleaving is called uniform interleaving. However, for performance reasons, turbo cords generally use non-uniform interleaving. Basic code means one or more efficiency codes of the above types. These can be, for example, a recursive systematic convolution code for a convolution turbo code, a humming or BCH block code for a block turbo code. Different types of concatenation can also be envisioned. In parallel concatenation, the same information is coded separately by each coder after being interleaved. In series concatenation, the output of each coder is interleaved and then encoded by the next coder. The number of basic coder used to realize this turbo code is called the turbo code dimension. One well-known turbo coding scheme consists of parallel concatenation of convolutional code type basic codes. This turbo code is indicated by the term PCCC. Examples of the series-connected turbo cord include a folding cord type basic cord and an SCCC using a block turbo cord using a block cord type basic cord.

【0013】

The information encoded by the turbo code can be decoded by an iterative method called turbo decoding. An example of turbo decoding is in the above document "Near Shannon Limit Error-Correcting Coding and Decoding: Turbo-codes" by C. Berrou, A. Glavieux, P. Thitimajshima, published in ICC-1993, in the minutes of the meeting. It is described on pages 1064 to 1070. This example relates to turbo decoding of a parallel connected turbo code. A number of basic decoders with weighted inputs and outputs are associated, and each decoder corresponds to the basic coder of the coding device. Weighted inputs and outputs are done with a probability, likelihood ratio, or log likelihood ratio called LLR. Weighted inputs and outputs are generally M-element symbols (M-ary) in the inputs and outputs of the base coder. Each of the symbol), eg, when a binary coder is used as the base coder, is associated with a bit. The decoders work one at a time in the case of series turbo decoding and at the same time in the case of parallel turbo decoding. An intermediate connection method can also be envisioned. Interleavers and de-interleavers allow each decoder to consider information items that appear in the same form as the corresponding coder's output form (and, in the case of parallel decoding, further input form). To do. Each basic decoder uses only a portion of all the information obtained at the input of the turbo decoder, i.e. a damaged version of the information at the output of the corresponding basic coder (and, in the case of parallel decoding, further information at the input). .. This information, called prior information, consists, on the one hand, of the information present at the output of the step prior to channel decoding (ie, generally the demodulator), and on the other hand, one or more previous basic decoders. It consists of the information generated by. Knowing this pre-information and the corresponding basic coder coding method, the basic decoder produces post-information, which is a more reliable estimate of the information at the input. Further information that is compared to the information at the input of the basic decoder is called external information. This external information is transmitted to the next basic decoder after interleaving or deinterleaving. Therefore, each decoder has prior information at the input and its quality is improved by the work of the previous basic decoder. This external information depends on the redundant information introduced by the corresponding coder. The method is repeated in that the external information calculated by the last one or more decoders in the series is propagated to the first one or more decoders in the series. The exchange of external information takes place only within the process and between the basic decoders from this process to the next. Therefore, each new step improves the reliability of the information produced at the output. After multiple iterations, this method will converge U. The threshold settings are applied to generate the decryption information.

【0014】

Needless to say, well-known turbo decoding includes, for example, various concatenated concatenation methods, depending on the type of turbo coding realized. For example, in turbo decoding corresponding to a series connected turbo code, the basic decoders are associated in the reverse order of the basic coder, and each basic decoder receives two preweighted information items, one of which is the corresponding basic. Corresponds to the output information of the coder, and the other corresponds to the input information of the corresponding basic coder. This basic decoder generates two post-weighting information items, one of which corresponds to the output of the corresponding basic coder, so that at the next iteration it will be the pre-input to the previous basic decoder after the corresponding interleaving. On the other hand, since it corresponds to the input of the corresponding basic coder, it becomes the pre-input of the next basic decoder after the corresponding deinterleaving at the same iteration.

【0015】

In any case, the external information is always defined as additional information provided by the basic decoding related to the basic coding for the prior information item and can occur at the input of the basic decoding.

【0016】

In addition, various types of algorithms can be used for basic decoders with weighted inputs and outputs. The basic decoder uses, for example, the MAP, LogMAP, and MaxLogMAP algorithms, also known as APP, LogAPP, and MaxLogAPP. All of these are obtained from the calculation of posterior probabilities, knowing the prior probabilities. For the description of such a decoding algorithm, for example, P. Robertson, P. Hoeher, E, published in European Trans.on Telecommun, Vol. 8, March-April 1997, pp. 119-125. .Villebrun's reference "Optimaland sub-optimal maximum a posteriori algorithms suitable for turbo You can refer to "decoding". The modified Viterbi algorithm type algorithm can also be used to relate each decision to a type of reliability measurement comparable to LLR (Log Likelihood Ratio). For example, the SOVA algorithm (soft output Viterbi algorithm) can be used. The block turbo code is described in the document "A very low complexity block turbo decoder for product codes" by R. Pyndiah, P. Combelles and P. Adde published in 1996 IEEE Globecom, pp. 101-105. As such, a chase algorithm can be used.

【0017】

For example, J. Hagenauer's article "Rate-Compatible Punctured Convolutional (RCPC) codes and their application" published in IEEE Trans., COM-36.4, 1988, pp. 389-400, or published in IEEE Trans. Also, the literature "New Rate Compatible Punctured Convolutional Codes for Viterbi" by LHC Lee Decoding, COM-42.2, 1994, pp. 3073-30079, can increase the efficiency of a code by a puncturing operation that consists of not transmitting a specific bit of information sequence. Is also known. These untransmitted bits are generally redundant information bits. This puncturing operation occurs at the transmission level after the coding operation. At the destination level, a mutual depuncturing operation is performed before the decoding operation. The puncturing behavior is defined by the puncturing scheme or matrix, and the depuncturing behavior is defined by the corresponding depuncturing scheme or matrix. Puncturing the redundant information bits reduces the ability of the code to correct and increases its efficiency.

【0018】

The error correction code of the turbo code family due to the state of the above technology maintains high enough efficiency, allows decoding operation with low complexity compared to the complexity of the code, and is very effective error. Allows acquisition of corrections. In that principle, turbo compounding is close to the best. However, performance close to that of the optimal decoder is obtained, and its complexity is clearly low. This is because the complexity of decrypting the basic code is high.

【0019】

It is known that the performance of the error correction code changes depending on the transmission conditions. The transmission condition means a parameter that particularly affects the transmission performance such as the signal-to-noise ratio. Performance is evaluated specifically by bit or packet error rate. In addition, many applications require a trade-off between performance on the one hand and usage parameters such as system complexity or decryption time on the other. In many cases, decoding does not compensate for all transmission errors. Access to the parameters that characterize the quality of the decrypted information has proven to be advantageous.

【0020】

Since turbo decoding consists of continuous decoding operations performed in a loop based on different redundant information each time, it is difficult to simply evaluate the quality of the decoded information. For this reason, many problems are presented.

【0021】

For example, it is difficult to predetermine the number of iterations that the decoding method must perform. Therefore, in general, the maximum number of iterations corresponding to the allowable complexity and / or maximum decoding time is determined, and the decoding process is interrupted if the latter appears to have converged to the transmission sequence before the maximum number of iterations. Stop criteria are used to do this. Detection of convergence can be done in different ways. For example, a CRC (Cyclic Redundancy Check) type error detection code can be used. During turbo decoding of a block, if the calculation of the error detection code shows that there are no more errors, the iterative decoding of the block is interrupted. One drawback of this first method is that the error detection code must be introduced at the time of transmission, which in particular reduces the overall efficiency of the channel coder. The other method consists of detecting stagnation in the output of the basic decoder during iterative decoding of the sequence. Such stagnation is effective in showing that repeated decoding does not further reduce the number of errors in the sequence being considered. Although this processing method makes it relatively easy to control the stoppage of the decoding process, information about the quality of the decoding sequence, i.e., the number of errors that the decoding sequence still has compared to the transmitted information, is Not given.

【0022】

For many applications, it is desirable to be able to know the number of errors remaining in the block of decrypted information. More generally, it can be advantageous to know the quality of the decrypted information at the output of each basic decoder and for each iteration.

【0023】

In the present invention, parameters that are easily accessible and that can evaluate the quality of the decoding sequence are sought.

【0024】

Based on the PCCC type turbo code by turbo decoding of the N bit sequence by the LogMAP type algorithm as an example, research in this direction has shown that, on the other hand, for the sequence of N external values in the output of the predetermined decoder during the predetermined iteration. There is a correlation between the average of the absolute values of the external information calculated in the above and, on the other hand, the number of errors remaining in the decoded N-bit sequence at the output of this given decoder for this given iteration. It became clear to do. The higher this average, the less likely it is that the number of errors remaining in the sequence will be small.

【0025】

It is possible to create a reference table showing the correspondence between this average and the average number of bit errors per sequence for different transmission conditions characterized by the signal-to-noise ratio.

【0026】

It is also possible to create a similar reference table from other quantities that show external information about a given number of information items in the output of the basic decoding operation, such as moments on the order of minimum, maximum, or greater than one.

【0027】

In the output of one or more basic decoding operations, from a certain number of weighted information items, the quality of the set of information items corresponding to the set of weighted information items in the output of the last basic decoding operation considered. The ability to calculate the quantitative characteristics of is more generally shown for all types of turbocodes in parallel or series connection.

【0028】

[Problems to be Solved by the Invention]

Therefore, it is an object of the present invention to provide a type of transmission method using error correction coding with a turbo code that provides parameters indicating the characteristics of decoded information quality in any decoding step during any decoding iteration. To make a suggestion.

【0029】

[Means for solving problems]

For this purpose, the present invention provides a coding process for generating a coded information item containing at least one redundant information item from a useful information item prior to the step of transmitting on the channel, and on the channel. After the step of transmitting, error correction including a decoding process for obtaining an estimate of the source information item from the encoded received information item by correcting the transmission error based on the at least one redundant information item. In a coded digital transmission method, the coding process is related to a plurality of interleaving steps and includes a plurality of basic coding steps acting in parallel or in series, and the decoding process is repeated and each. For the iteration, it comprises a plurality of basic decoding steps corresponding to the plurality of basic coding steps associated with the plurality of applied interleaving and deinterleaving steps, each of the basic decoding steps being one. Generate at least one weighted output information item that can be transmitted to or more other basic decoding steps, the method of which is at least one from a set of weighted output information items generated by at least one basic decoding step. The at least one characteristic quantity and at least one configuration of the characteristic quantity determination step for calculating the characteristic quantity and the decoding information quality parameters related to the decoding information item set corresponding to the weighted output information item set. We propose a method characterized by including a decoding information quality parameter determination step for determining from parameters.

【0030】

As defined above, the present invention applies to all types of transmission methods using turbo-encoded error correction coding, regardless of series or parallel turbo coding.

【0031】

The present invention is also applied within the range of turbo equalization, where the channel is considered a convolution coder, or within the range of turbo detection.

【0032】

The decrypted information quality parameters can also be used after turbo decoding or within the turbo decoding process.

【0033】

According to another aspect of the present invention, each of the basic decoding steps uses a part of the received information, and a part of the received information includes an external information item that can be transmitted to another basic decoding step. Corresponding to an information item containing redundant information related to the corresponding basic coding step for generating an output information item, at least one external information item obtained in one iteration is transmitted to the other iteration, said. The characteristic quantity determining step calculates at least one characteristic quantity from a set of external information items in the output of the basic decoding step during the basic decoding step.

【0034】

Advantageously, the characteristic quantity is a statistic. This can be the average of the absolute values of external information calculated for the set of external information items. It can also be other statistics such as variance, minimum, or maximum.

【0035】

According to another aspect of the present invention, the quality parameter determination step sets the decoding information quality parameter to a characteristic quantity calculated during the basic decoding step from a set of weighted output information items in the basic decoding step. The decoding is determined from the set of weighted output information items corresponding to the set of weighted output information items in the basic decoding step, other characteristic quantities calculated during the previous basic decoding step, and at least one configuration parameter. The information quality parameter relates to a set of decoding information items corresponding to the set of weighted output information items in the basic decoding step.

【0036】

Advantageously, the quality parameter determination step determines the quality parameter from the characteristic quantities calculated during the basic decoding step corresponding to the final basic decoding step of the decoding process.

【0037】

However, the quality determination step can also determine the quality parameter from a single characteristic quantity calculated during the final basic decoding step of the decoding process.

【0038】

In certain applications, the output information quality parameter is an integer indicating the estimated number of errors in the block under consideration.

【0039】

In other applications, the output information quality parameter is a scalar used as a weighting factor.

【0040】

In this case, the larger the parameter, the more reliable the output information.

【0041】

The configuration parameters can be parameters that characterize the transmission conditions, eg, the signal-to-noise ratio.

【0042】

The information quality parameter determination step may use a predetermined algorithm that allows the information quality parameter to be calculated as a function of one or more characteristic quantities and constituent parameters.

【0043】

This information quality parameter determination step also uses a predetermined reference table that allows the information quality parameter to be selected as a function of one or more characteristic quantities and constituent parameters.

【0044】

According to another aspect of the invention, the received information is processed by an N-bit decoding sequence, the set of decoding information items being a sequence of binary information items containing N bits.

【0045】

This embodiment of the present invention is particularly advantageous for applications comprising assigning the decoding information quality parameters to the decoding sequence in the output of the decoding process.

【0046】

Alternatively, the received information is processed by an N-bit decoding sequence, and the set of decoding information items is a sequence of binary information items showing only a portion of the N-bit decoding sequence.

【0047】

In the present embodiment of the present invention, the decoding information quality parameter corresponds to the decoding sequence portion in order to favor a particular set of external information items to a greater or lesser extent and improve the convergence of the turbo decoding method. Converting a set of external information items to a weighting factor is advantageous for applications that consist of use in the turbo decoding process itself.

【0048】

According to another aspect of the invention, the basic decoding step has inputs and outputs weighted in terms of probability, likelihood ratio, or log likelihood ratio.

【0049】

According to another aspect of the invention, the coding process comprises at least one puncturing step and the decoding process comprises at least one corresponding depuncturing step.

【0050】

According to another aspect of the present invention, in a combination of transmission methods using a number of decoding processes associated with the exact same coding process, the decoding information quality parameters obtained for each of the decoding processes are decoded. It can be used as a weighting factor for a set of information items, considering the combination of these sets.

【0051】

According to another aspect of the present invention, the decoded information quality parameter can be used as a parameter of the joint detection step in a transmission method further including a joint detection step.

【0052】

The above and other features of the invention will be clarified by reading the description of the exemplary embodiments below, which description is given in connection with the accompanying drawings.

【0053】

BEST MODE FOR CARRYING OUT THE INVENTION

Embodiment 1. The present invention will be exemplified, in particular, with respect to a transmission method of a type by error correction coding using a two-dimensional PCCC type turbo code. FIG. 1 schematically shows a turbo coding device to which this method is applied, and FIG. 2 shows a turbo decoding device to which this method is applied. The present invention can be easily generalized to larger dimension turbo cords using other types of basic cords and / or by different coupling schemes, especially serial coupling schemes.

【0054】

Mainly, the error correction coded digital transmission method using the PCCC turbo code includes a coding process before the transmission process and a decoding process after the transmission process.

【0055】

The coding process will be illustrated using the coding device schematically shown in FIG.

【0056】

Primarily, the coding device 10 has two basic coders 11 and 12, with a size N interleaver 13 acting between them.

【0057】

Each of the basic coder 11 and 12 is a coder (coder) that uses an iterative systematic convolution (RSC) code. As is well known, each of these basic coder uses a series of auxiliary information items stored in a shift register type device, and these items are a useful information item and a pre-computed auxiliary information item, respectively. Obtained by a mathematical combination of. In the example presented here, shift register 23 stores auxiliary data calculated by the exclusive OR gate 21. The convolution product is realized by the exclusive OR gate 22, and the generator polynomial is formed by the output of register 23. Useful information is systematically transmitted by the redundant information item side that appears in the output of the exclusive OR gate 22. In step 13 of interleaving size N, the order in which the data is considered by each of the base codes is changed. Thus, each of the coder 11 and 12 produces the associated redundant information item. Source information is sent only once. Therefore, the coding information that appears by the coding process is block 20, which includes useful information or a systematic part, and a part of the coding information corresponding to each of the two redundant information items or the basic code. Needless to say, the two basic codes can be different. After multiplexing 14, the coded information can be punctured 15. The efficiency of each of the basic coder is 1/2, and the efficiency of the turbo code is 1/3 because the systematic part is transmitted only once. Needless to say, this efficiency is increased by puncturing. Therefore, by puncturing half of the redundant bits of each base code, half the efficiency is obtained.

【0058】

The encoded and possibly punctured information is transmitted by sequence 20 of length N / R (where R is the efficiency of the turbo code), including puncturing. These sequences are modified by channel. These sequences are then received by the decryption device, perhaps after depuncturing. The weighting information sequence 30 of length N / R appears at the input of the demultiplexer 31. Each of these blocks 30 constitutes a received information item to be decrypted. These blocks include a received information unit called systematic information X, which corresponds to useful information, a first received information unit, which corresponds to redundant information of the first basic code called first redundant information Y1, and a first unit. 2 Redundant information Includes a second received information section corresponding to the redundant information of the second basic code called Y2. The decoding process operates by an N-bit decoding sequence corresponding to the received N-block sequence.

【0059】

The decoding device has a basic decoder 32 corresponding to the first basic coder 11 and a basic decoder 33 corresponding to the second basic coder 12 connected in series. In the example considered here, the basic decoder using the LogMAP type algorithm has weighted inputs and outputs in the form of a log likelihood ratio (LLR). Therefore, the external information is initialized to 0, and the systematic information and the external information are combined by addition. Needless to say, if the amount that the basic decoder can handle is something else, it needs to be changed accordingly. For example, if these are likelihood ratios, the external information is initialized to 1 and the combination is done by product. If these are probabilities, the external information is initialized to 0.5 and the combination is also a product.

【0060】

The reception information unit X corresponding to the source information is further transmitted to the basic decoders 32 and 33. On the basic decoder 32 side, the adder 37 is an external information item e2._k_’_-1Is added to this systematic information X. On the basic decoder 33 side, the adder 39 is an external information item el interleaved by an interleaver 35 of size N corresponding to the interleaver 13._k_’Is added to the systematic information X interleaved by the interleaver 34 of size N corresponding to the interleaver 13. Further, the reception information unit Y1 corresponding to the redundant information of the first basic code is transmitted to the decoder 32, and the reception information unit Y2 corresponding to the redundant information of the second basic code is transmitted to the decoder 33.

【0061】

The exponent k'indicates the current iteration of the decoding process, so the external information e2_k_’_-1Is external information el_k_’Is calculated in the iteration before the iteration in which is calculated.

【0062】

External information el_k_’Is systematic information X and external information e2_k_’_-1Is obtained with the output of the basic decoder 32 during the iteration k'by subtracting with the subtractor 38.

【0063】

External information e2_k_’In the subtractor 40, interleaved systematic information X'and interleaved external information e'l._k_’By subtracting, in the output of the base decoder 33 during the iteration k', the interleaved form e'2_k_’Obtained at. It is deinterleaved by a size N deinterleaver 36 that corresponds to the interleaver 13 before being sent to the next iteration.

【0064】

At the end of the decoding process, i.e. after a predetermined number of iterations, the decoding sequence at the output of the second basic decoder 33 is deinterleaved and analyzed by the decision block 41 to form the decoding sequence.

【0065】

First, external information e2₀Is initialized to 0. At the first iteration, the systematic information X forms the pre-input information of the first basic decoder 32. From the first redundant information Y1 by the first basic decoding, the weighted output information item D1₁Is generated. Weighted output information item D1₁Corresponds to the first decoding sequence estimation, systematic information and external information el₁The latter is written in combination with (corresponding to the reliability improvement associated with the first basic decoding). This makes D1₁= X + el₁Is obtained, external information el₁Is written as the difference between the weighted output information of the first decoder (here, the log likelihood ratio at the output) and the weighted input information of the first decoder (here, the log likelihood ratio at the input). .. This external information el₁Is added to the interleaved and interleaved systematic information X'to form the pre-input information of the second basic decoder 33. Weighted output information item D'2 by the second basic decoding₁Is generated from the second redundancy information Y2. Weighted output information item D'2₁Corresponds to the second decoding sequence estimation, interleaved systematic information and interleaved external information e'l₁And interleaved external information e'2₁The latter is written in combination with (corresponding to the reliability improvement associated with the second basic decoding). This makes D'2₁= X'+ e'1₁+ e'2₁Was obtained and interleaved external information e'2₁Is expressed as the difference between the weighted output information of the second decoder (here, the log likelihood ratio at the output) and the weighted input information of the second decoder (here, the log likelihood ratio at the input). .. Interleaved external information e'2₁After deinterleaving, external information e2₁To form. External information e2₁Is added to the systematic information X to form the pre-input information of the first basic decoder 32 for the second iteration. In basic decoding, the first redundant information Y1 to the weighted output information item D1₂Is generated. Weighted output information item D1₂Corresponds to a new more reliable decoding sequence estimation. New external information item related to decoder 32 el₂Is interleaved and added to the interleaved systematic information X'to form the pre-input information of the second basic decoder 33. Weighted output information item D'2 from the second redundancy information Y2 by the second basic decoding₂Is generated. Weighted output information item D'2₂Corresponds to a new more reliable decoding sequence estimation. New external information item related to decoder 33 e2₂Is added to the systematic information X to form the pre-input information of the first basic decoder 32 for the third iteration. The process then continues in the same way, and the external information gains reliability (ie, the amplitude in this case represented by the log likelihood ratio as the iteration progresses). Weighted output information item D2 in the output of the second basic decoder 33 at the end of the iteration of a given number k_iThe interleaved decoding sequence composed of is deinterleaved, reaches the limit, and produces a decoding sequence.

【0066】

FIG. 5 shows an embodiment of the present invention applied in the case of an i-dimensional PCCC such as the above-mentioned two-dimensional PCCC. According to the present invention, in the characteristic quantity determination step 86, during the final basic decoding step 83 of the decoding process, that is, the step corresponding to the i-th basic decoding of the k-th iteration, i in the k-th iteration. Mean of absolute values of external information calculated for a sequence of N external values in the output of the second decoder E | ei_kAn algorithm consisting of computing | is executed. In the case of the 2D PCCC described above, in this step 86, in the basic decoding performed by the second decoder 33 during the kth iteration, the N external value at the output of the second decoder 33 during the kth iteration. Mean of absolute values of external information calculated for the sequence E | e2_k| Is calculated.

【0067】

In the decoded information quality parameter determination step 85, the characteristic quantity E | ei_k| Is received. The process also receives other parameters that are constituent parameters, i.e. parameters that can characterize the system in which the decrypted information quality is evaluated. These can be, in particular, the estimated signal-to-noise ratio, the size N of the decoding sequence, the value i, and the value k. These can also be the parameters that characterize puncture ringing, or any other parameter that can change the quality of the decrypted information.

【0068】

At the end of basic decoding step 83, characteristic quantity E | ei_kIn order to determine the decoding information quality parameters associated with the sequence of N decoding information items corresponding to the sequence of N external values in the output of the i-th basic decoder from | and the configuration parameters, a predetermined algorithm is used in the determination process. Alternatively, a predetermined standard table is used. This algorithm or a reference table thereof has been predetermined as a function of the average number of errors per decoding sequence for different constituent parameters by conventional studies on the mean distribution of absolute values of external information.

【0069】

As an example, FIG. 3 graphically illustrates the results of such studies on the PCCC of this embodiment. This is the average of the absolute values of the external information in the output of the second decoder of the tenth iteration E | e2₁₀The distribution of | is shown as a function of the average number of errors for each decoding sequence, and the length of the sequence (corresponding to the size of the interleaver) is N = 640. This distribution is created for a signal-to-noise ratio of three values, the first line corresponds to a signal-to-noise ratio of 0 dB, and the second line has a signal-to-noise ratio of 0.25 dB. The third line corresponds to the case where the signal-to-noise ratio is 0.5 dB. The signal at the output of the turbocoder is modulated by BPSK modulation and transmitted over the channel with Additive White Gaussian Noise (AWGN). Quantity E | e2₁₀| Is calculated for a sufficient sequence to give a "smooth" distribution. On the x-axis, the number of errors is shown at 10 intervals so that each interval has sufficient statistics. When dealing with 0 errors, they are considered separately. Because the corresponding E | e2₁₀This is because the value of | is much higher than in other cases. On the y-axis, the absolute value of external information is the log (E | e2)₁₀|) Is shown.

【0070】

For example, the distribution graph thus obtained, used by the reference table, is derived from the characteristic quantities associated with the sequence of N-decoded data items for a given signal-to-noise ratio forming the constituent parameters. Allows deduction of the estimated number of errors remaining in. Here, this number of errors forms an information quality parameter, which, according to the invention, is related to the decoding information sequence. For example, for a signal-to-noise ratio of 0 dB, the characteristic quantity E | e2 of 2.51 with a logarithm of 0.4.₁₀A value of | indicates that about 30 errors remain in the 640 decryption bits. This quality parameter can be of a different type depending on the application.

【0071】

Needless to say, the information quality parameter determination step can be based on a large number of characteristic quantity distributions corresponding to various constituent parameter values as shown in FIG.

【0072】

FIG. 4 is a flow diagram which more generally illustrates the basic principle of the decoding process using the determination of the decoding information quality parameter according to one embodiment of the present invention.

【0073】

According to the present invention, in the characteristic quantity determination step 54, the i'th basic decoding of the iterative k', the (i-1) th basic decoding of the iterative k', and the i-th basic decoding of the iterative k'are made respectively. During the basic decoding steps of the decoding process, such as the corresponding illustrated steps 51, 52 and 53, the weighted information items corresponding to all or part of the decoding sequence in each output of these basic decoding steps. aj₁、_{j = 1 ... i}_、_{l = 1 ..}_.kFunction f (aj_l)_{j = 1 ... i}_、_{l = 1 ... k}Executes an algorithm consisting of calculating. The characteristic quantity can be a statistic such as an amplitude average, a variance, a minimum value, and a maximum value. The function f can give a characteristic quantity of a set of weighted information items in the output of the final basic decoding step of the decoding process. It is also a characteristic quantity of a set of weighted information items in the output of the final basic decoding step of the decoding process, regardless of whether it is located at the end of the decoding iteration or within the decoding iteration. Can be given.

【0074】

In the decoded information quality parameter determination step 55, the characteristic quantity f (aj)_l)_{j = 1 ... i}_、_{l = 1}_{... k}And the configuration parameters are received. Here again, these are, among other things, the estimated signal-to-noise ratio, the size N of the decoding sequence, the values i and the value k, the parameters that characterize one puncture ring or multiple puncture rings performed, and one inter It can be a parameter or the like that characterizes leaving or multiple interleavings.

【0075】

At the end of the basic decoding step 53, the characteristic quantity f (aj)_l)_{j = 1 ... i}_、_{l = 1 ... k}And from the configuration parameters, the weighted information item ai in the output of the i-th basic decoder during the k-th iteration._kIn order to determine the decrypted information quality parameters associated with the set of decrypted information items corresponding to the set of, the information quality parameter determination step uses a predetermined algorithm or a predetermined reference table. This algorithm or these reference tables have been pre-determined as a function of the decoded information quality parameters for different constituent parameters by conventional studies on the distribution of characteristic quantities. This quality parameter can be an integer indicating the estimated number of errors in the set of decryption information items, a scalar used as a weighting factor, or any other parameter required for a given application.

【0076】

In the above embodiments, the decoded information quality parameters according to the present invention are specifically designed for applications in the downstream sector of turbo decoding processing, examples of which will be described subsequently. In this case, the parameters are generally calculated for a set of decoding information items that correspond to the entire decoding sequence. Further, the parameters are generally calculated during the basic decoding step, which is the last step in the decoding process, or during the final basic decoding step in the decoding process. However, such decryption information quality parameters can also be used within the turbo decoding process. The set of decrypted information items associated with the decrypted information quality parameter means any set of decrypted information items at the end of the basic decoding process, not necessarily the final step of the decoding process. For example, such decoding information quality parameters can be calculated at the end of the basic decoding step to change the next basic decoding step. In this case, this parameter also corresponds to the decoding sequence portion, for example, to give different weights to a particular subset of the decoding sequence and to favor the subset according to the quality of information contained therein during the next step. Can be calculated for a set of decryption information items.

【0077】

The present invention relates to all types of transmission methods using turbocodes, regardless of series or parallel turbo coding, using turbocoded error correction coding or a hybrid coupling scheme that mixes series and parallel coupling. Applies.

【0078】

An embodiment of the present invention applied to a series turbo cord will be briefly described.

【0079】

For example, in a series-connected turbo code, each basic coder generates a basic coding information item from the basic coding information items generated from the previous basic coder, and the i basic coder is generated by the (i-1) interleaver. It is separated. The puncturing process is distributed in the coding process, and the output of the jth coder is probably punctured by the jth puncturing vector and jth before being encoded by the (j + 1) th coder. Interleaved by the interleaver of. The size of each interleaver depends on the efficiency of the previous coding process, especially after puncturing the previous base coder.

【0080】

In the turbo coding example corresponding to the series connected turbo coding above, the basic decoders are associated in the reverse order of the basic coder, and each basic decoding step receives two preweighted information items and one. Is called the first input information corresponding to the output information of the corresponding basic coder, and the other one is called the second input information corresponding to the input information of the corresponding basic coder. This basic decoding step produces two post-weighted information items, one is called the first output information that corresponds to the output of the corresponding basic coder, and thus interleaving and corresponding in the next iteration. After puncturing, it becomes the pre-input of the previous basic decoder, and the other one is called the second output information that corresponds to the input of the corresponding basic coder, thus deinterleaving and corresponding in the same iteration. After depuncturing, it becomes the pre-input of the next basic decoder. The first input information corresponds to the information decoded by the basic decoding step. The second output information corresponds to the information decoded by the basic decoding step, and is composed of a combination of the second input information and the external information item.

【0081】

According to the present invention, in the characteristic quantity determining step, in the basic decoding step, the characteristic quantity is calculated from the set of weighted information items in the output of the basic decoding step. For example, referring to the turbo decoding example described above, the characteristic quantity is calculated from a second set of output information values corresponding to the decoding sequence or the decoding sequence portion. In the decoding information quality parameter determination step, the decoding information quality parameters related to the set of decoding information items corresponding to the set of weighted information items are determined from the characteristic quantity and the configuration parameters. The decrypted information item means an information item that is decoded by the basic decoding step considered. This will be the decoding information corresponding to the useful information if the considered basic decoding step is the final step of the decoding process.

【0082】

The present invention is applied, for example, to a weighted combination method of frames transmitted and decoded differently, which is the same at the time of transmission, for example, a soft handover type method.

【0083】

FIG. 6 schematically shows the basic principle when the present invention is applied to the weighting combination method of frames. This method relates to a combination of transmission methods using a number of decoding processes associated with the exact same coding process. The sequence encoded by the turbocoded coding process 60 is subjected to an iterative step 61, so that the encoded sequence is transmitted to channels 62 with different n. Here, channel means, among other things, a channel in a broad sense that exhibits the effects of puncturing, modulation, the physical channel itself, demodulation, and detection. Each sequence modified by the corresponding channel is subjected to its own turbo-decoding iterative decoding 63. The result is a decoding sequence of n. By applying the present invention to each of the decoding processes, it becomes possible to associate each of these n sequences with each weighting factor determined by the decoding information quality parameter determination step. In combination step 64, the n sequences are combined by applying their respective weighting factors to each sequence in order to obtain a final decoding sequence that represents the source information sequence. Symbol α_kIs the decryption sequence d_kGiven to the weighting factor associated with (k = 1 ... n) and the symbol d given to the sequence obtained by the weighting combination, d = f (α)_k, D_k)_{k = 1 ... n}Is obtained.

【0084】

The above application of the present invention is particularly advantageous in the field of mobile phones, for example, when a mobile phone communicates with a large number of base stations.

【0085】

The present invention is also applied, for example, in a transmission method including a joint detection step. FIG. 7 schematically shows the basic principle when the present invention is applied to the iterative decoding method using the joint detection block. The joint detection block is the second best detection block of controllable efficiency and complexity. The corresponding joint detection step 70 receives, for example, a sequence of weighted bits damaged by ISI (intersymbol interference) or MAI (multiple access interference) and sends them to the turbo-decoding iterative decoding process 71. Convert to the corresponding weighted bit sequence. At the end of the decoding process, a quality indicator corresponding to the decoding sequence is generated in the decoding information quality parameter determination step. In switching step 72, the quality indicator is compared to a given threshold, and if the quality is not satisfactory, the sequence of weighted bits is returned to detection step 70, while at the same time the complexity of the detection block by the quality indicator. And new settings for efficiency are instructed.

【0086】

[Effect of the invention]

As described above, according to the present invention, during any decoding iteration, in any decoding step, a type of transmission using error correction coding with a turbo code that provides parameters indicating the characteristics of the decoding information quality. The method is provided.

[Simple explanation of drawings]

[Figure 1]

It is a block diagram which showed the 2D PCCC type turbo code coding device.

[Figure 2]

FIG. 5 is a block diagram showing a series mode decoding device associated with the coding device of FIG.

[Fig. 3]

It is a figure which showed the graph which shows the average distribution of the absolute value of the external information as a function of the average number of errors for every decoding sequence for a predetermined decoding process and different transmission conditions.

[Fig. 4]

It is a flow chart which shows the basic principle of the decoding process using the determination of the decoding information quality parameter by one Embodiment of this invention.

[Fig. 5]

It is a flow chart which shows the basic principle of the decoding process using the determination of the decoding information quality parameter by another embodiment of this invention.

[Fig. 6]

It is a flow chart which shows the application of this invention to the transmission method including a plurality of decoding processing which combines output.

[Fig. 7]

It is a flow chart which shows the application of this invention to the transmission method including a joint detection step.

[Explanation of symbols]

10 coding devices, 11,12 basic coder, 13 interleaver, 14 multiplexing, 15 puncturing, 21,22 exclusive OR gate, 23 registers, 30 weighted information sequence, 31 demultiplexer, 32,33 basic decoder, 34,35 interleaver, 36 deinterleaver, 37,39 adder, 38,40 subtractor, 41 decision block, 50 received information item, 51,52,53 basic decoding, 54,86 characteristic quantity determination, 55, 85 Decoding Information Quality Parameter Determination, 60 Coding, 61 Iterations, 62 Channels, 63 Decoding, 64 Combinations, 70 Detections, 71 Iterative Decoding, 72 Switching, 83 Last Basic Decoding.