RU2337478C2 - Decoding of highly excessive code with parity check, using multithreshold message transfer - Google Patents

Abstract

FIELD: physics, communications.

SUBSTANCE: multithreshold circuit for iterative decoding of accepted highly-excessive code word with parity check includes comparison of updated bit safety to threshold of accepted code word recovery. In each iteration, bit safety and recovered code word based on comparison involving threshold updated for this iteration are updated. Invention embodiment versions involve methods of decoding and/or relevant coding and device using threshold taking two or more values during iterative coding.

EFFECT: accuracy and good response at lower iteration number.

46 cl, 6 dwg

Description

Technical field

Embodiments of the invention relate generally to error correction in communication signals.

State of the art

A communication channel, whether it is a fiber-optic channel, a coaxial channel, a wired channel, a wireless channel or a bus connecting system elements through which large amounts of data are sent, can introduce noise and errors into the information transmitted through such a channel. To correct errors added by the communication channel, information can be sent in the form of code words, where each code word contains the same total number of bits and in which a certain number of bits are information bits (message bits), and several bits are used to correct errors. A codeword of length n bits includes k bits for the length of the message code and r = n-k redundant bits. r bits are for error correction and may be r parity bits. The parity check matrix H contains a set of parity control equations that define a codeword according to the relation:

HC ^T = 0,

where C is the n-dimensional vector of bits of the codeword. If this ratio is not satisfied at the receiver, then the received codeword is invalid and must either be corrected or retransmitted.

There are many different schemes for decoding a codeword that has passed through a communication channel. Some circuits can provide accuracy, while other circuits can provide fast decoding. For high-speed data transmission, a decoding scheme is needed to ensure accuracy and speed of operation, but at the same time, its implementation should not be very complex.

Brief Description of the Drawings

Figure 1 - block diagram of a variant of the method for iteratively decoding information obtained from a communication channel, according to the present invention;

FIG. 2 is an embodiment of an apparatus for iteratively decoding information obtained from a communication channel according to the present invention; FIG.

Figure 3 is a flowchart of an embodiment of a method for encoding information for transmitting a codeword representing information to a communication channel according to the present invention;

FIG. 4 is an embodiment of an apparatus for encoding information that is iteratively decoded at a receiving end of a communication channel using a plurality of threshold values according to the present invention; FIG.

5 is a block diagram of an embodiment of a communication network having a system for encoding a high redundancy codeword with parity and a system for decoding a version of a coded high excess codeword with parity obtained from a communication channel between two systems according to the present invention;

FIG. 6 is a flowchart of an embodiment of a method for iteratively decoding a high-redundancy parity codeword using a decoder that can be configured as shown in FIG. 5 according to the present invention.

DETAILED DESCRIPTION OF THE INVENTION

In the following detailed description, reference is made to the drawings, where, as an illustration, possible specific details and embodiments of the invention are shown. These options are described in sufficient detail to enable those skilled in the art to practice the present invention. You can use other options and make structural and logical changes, as well as changes in electrical circuits, without changing the scope of the invention. The disclosed embodiments are not necessarily mutually exclusive; on the contrary, some disclosed embodiments may be combined with one or more other disclosed embodiments to create new variations. Thus, the following detailed description should not be construed as limiting, and the scope of embodiments of the present invention is determined only by the attached claims, together with the full scope of equivalents to which the claims claim.

High redundancy parity check (LDPC) codes provide an efficient direct error correction tool in high-speed communication systems due to the low decoding complexity and low bit error rate (BER) achieved in an additive white Gaussian noise (AWGN) channel. Standard LDPC decoders operate using a single-threshold decoding scheme. Examples of well-known LDPC decoders include, among others, a minimum sum algorithm, a posteriori probability decoding (APP) algorithm, and a uniformly most powerful (UMP) decoder.

LDPC codes use a parity matrix that contains most zeros and a limited number of units. The binary (n, γ, ρ) LDPC code has a codeword or block length of n bits and a parity matrix that contains exactly γ units in each column and exactly ρ units in each row. In the LDPC code, each code bit is checked with γ parity checks, and each parity check uses ρ code bits. In addition, the code has a speed R, defined as R = k / n, where k is the number of message bits (information bits) in the codeword that has n bits.

The iterative decoding process may include checking whether or not the parity equation is satisfied for each bit received from the channel. If all the parity equations are satisfied, then the received bits are considered to be the transmitted codeword. If most parity equations for a given received bit are not satisfied, then the value of this received bit can be reversed; otherwise, the value of this received bit is stored. Changing the value of a bit to the opposite means changing the zero by one or units by zero. After recovering the codeword with changing the bit value to the opposite, the parity equations are checked again. If all the parity equations are satisfied, then it is considered that the received bits are the transmitted codeword; otherwise, the iterative procedure continues until the maximum number of valid iterations is reached.

For LDPC decoding, a confidence distribution decoding algorithm based on the use of probabilistic values can be applied. As a rule, iterative decoding of the LDPC code is associated with the determination of the vector x with the maximum probability, where Hx ^T mod 2 = 0. The probability x can be given by the product of the probabilities that for each bit position n, X _n = x. For a binary codeword, x takes the values 0 and 1, so the probability that x = 1 is equal to one minus the probability that x = 0. In addition, probabilistic values can be represented by logarithmic likelihood ratios. For a binary signal with probability p defined as probability 1, the log-likelihood ratio (LLR) can be given as

LLR (p) = log _e ((1-p) / p).

When decoding sum-products, decoding is performed using LLR values.

As for the process associated with the opposite change in the value of the received bit, the initial hard decision was made based on the signal received from the communication channel. A tough decision is made for a bit by separately evaluating that bit. With a tough decision for a binary bit, the decoder sets the bit to 1 or 0 if that bit is less than or greater than the fixed value set by the decoder. If the decoder sets the fixed value to zero, with a hard decision, the bit is set to 1 if the received signal for this bit is positive, and set to 0 if the received signal is negative. Alternatively, with a hard decision, the bit can be set to 1 if the received signal for this bit is negative, and set to 0 if the received signal is positive.

A soft decision is a decision in which the probability that a bit will represent one or zero is taken into account when determining the value of a bit based on the received signal. Soft decision decoders can use information related to the degree of confidence in the received signal (“soft” received information) and channel property information to provide probabilistic expressions for the received signal associated with the transmitted signal. While a hard decision for a binary process provides a hard decision vector where the vector positions or vector bits are equal to one or zero, a soft decision can provide a hard decision vector and a vector of reliability or bit reliability. The binary vector of a hard decision provided by the soft decision is a vector consisting of ones and zeros, where each bit of the vector is created individually in accordance with a fixed criterion. The reliability vector includes a set of bit reliability, with each bit reliability corresponding to a bit of the corresponding hard decision vector. Bit reliability provides a measure of confidence in the corresponding bit in the hard decision vector. Bit reliability can be expressed as probability, a logarithmic likelihood ratio, or other weighted coefficient correlated with the communication channel over which the received codeword is transmitted. Bit reliability can be directly correlated with the magnitude of the signal representing the received bit.

As is known to those skilled in the art, various decoding schemes implement soft decoding schemes using probabilities. According to one scheme, a posterior probability is calculated for each bit of a code word in soft decision processes, which is the probability that a given bit is equal to one, provided that all restrictions for parity are satisfied. The intrinsic or a priori probability is a bit probability that does not depend on code constraints, and the "improper" probability takes into account what has become known. For example, in the sum-product algorithm for decoding, the logarithmic likelihood ratio (LLR) of a received signal representing one bit is set based on the channel information. For a channel with AWGN type noise and SNR, the initial LLR for this bit can be set as

, где n - длина кода, k - длина сообщения, а SNR - это SNR на один бит сообщения. LLR_i функционально относится к надежности R_i бита i. Надежность - это мера доверия к значению бита. Итеративное декодирование в различных схемах включает в себя обновление надежности, соответствующей принятым битам, и/или обновление надежности, соответствующей ненулевым элементам матрицы H контроля четности. Указанные итеративные процессы включают в себя проверку достоверности обновленного тестового кодового слова посредством соотношения Hx^T = 0. Если тест не дает достоверное кодовое слово, то итеративный процесс продолжается, пока не будет получено достоверное кодовое слово или пока не будет выполнено максимальное количество итераций.where Y _i is the received signal and σ is the standard deviation of AWGN. The standard deviation can be defined through the SNR for AWGN as

where n is the code length, k is the message length, and SNR is the SNR per bit of the message. LLR _i functionally refers to the reliability of R _i bits i. Reliability is a measure of confidence in the value of a bit. Iterative decoding in various schemes includes updating the reliability corresponding to the received bits, and / or updating the reliability corresponding to non-zero elements of the parity check matrix H. These iterative processes include verifying the validity of the updated test codeword using the relation Hx ^T = 0. If the test does not give a valid codeword, then the iterative process continues until a valid codeword is received or until the maximum number of iterations has been completed.

In one embodiment, a multi-threshold decoder is designed to provide an iterative decoding algorithm with reduced complexity and good error correction characteristics for decoding LDPC codes from a channel with a soft output. This option for a multi-threshold decoder with reliability calculation for fast LDPC decoding works using a channel with a “soft” output (soft decision) to create a hard decision vector and a reliability vector of received bits. Bit reliability is used to calculate the reliability for parity, and the overall reliability of the parity is set to the minimum value of bit reliability for all bits that are included in this parity. To recalculate bit reliability, a weighted voting procedure is used for parity, where the weights are the reliability of the parity. In these options, if the value of the weight of the parity is reduced, the quality of decoding increases. Initially, bit reliability is assigned a value of bit reliability correlated with the value of the received bit. Parity voting applies to the current hard decision vector. Based on the voting for parity compared to the threshold value in this iteration, the bit reliability is set equal to the newly calculated bit reliability, which is increased by the value of the parity reliability; otherwise, the recalculated bit reliability is reduced by the parity reliability value.

Threshold values are provided for each iteration, and these threshold values can be calculated during one or more iterations. Thresholds are used to make decisions about whether to update the bit and its associated reliability. If a specific reliability is less than a threshold value, then the hard decision bit associated with this reliability changes its value, and this new value is assigned to the absolute value of the newly calculated reliability. This decoding scheme is iterative, and the thresholds can vary from one iteration to another, and at least two threshold values are provided or planned for calculation during the decoding process. Although thresholds can be calculated during the iteration, their values can also be determined in advance. In another embodiment, at the first iteration, the decision threshold by mistake is set equal to the minimum of bit reliability, and then at all other iterations the decision threshold is mistakenly set to zero. In this embodiment, the multi-threshold decoding scheme is a decoding scheme using two thresholds. The two-threshold scheme provides an increase in the error frequency of bits from 0.1 to 0.4 dB compared to fast decoding without using a threshold with many values.

An option for a new circuit that can be implemented in LDPC decoders includes a multi-threshold decision circuit. Using a multi-threshold decision scheme means that the threshold value changes to a calculated value or a predetermined value during one or more decoding iterations. In one embodiment, during the first iteration, the maximum threshold value (absolute value) is used, and in subsequent iterations, the threshold changes downward. Changing the threshold values allows you to use a high threshold value during the first iteration, which eliminates or reduces the spread of errors. Thus, if a decoder corrects even a small number of errors, decoding at subsequent iterations is greatly facilitated. Using a variant with a multi-threshold decoding scheme may lead to a decrease in the number of iterations performed during decoding. These options can be implemented with a bit error rate during decoding close to the decoding scheme based on maximum posterior probability.

In one embodiment, a multi-threshold decoder provides an iterative decoder using a soft decision input vector to create a hard decision and a corresponding reliability vector that describes the absolute likelihood value for each bit. At each iteration, the decoder calculates the reliability of each parity equation using bit reliability. Then, for each bit and parity check, if the parity equation is not satisfied, the corresponding bit reliability (which corresponds to the bit or parity equation) is reduced; otherwise, this bit reliability is increased. When all equations are checked, a tough decision is made using the following rule: if the updated bit reliability is greater than the threshold value, then the bit and its reliability remain unchanged; otherwise, the bit value is reversed (inverted), and its reliability is changed in the function of the updated reliability and threshold value. If the updated reliability is less than 0 but greater than the threshold value in this iteration, then the new bit reliability value is set to 0. Updating the threshold value is planned at least once during the decoding process. The process is repeated until the codeword is restored, or until the decoder performs the maximum allowable number of iterations.

In FIG. 1 shows a flowchart of an embodiment of a method for iteratively decoding information received from a communication channel. At 110, a codeword is received. The codeword may contain a number of information bits or message bits and several bits for use in detecting errors and / or correcting them. In step 120, a threshold value is selected. In one embodiment, a threshold value is selected using bit reliabilities correlated with the received codeword during each iteration. Such a choice can be made by calculating the minimum reliability for comparison, originally formed on the basis of the soft decision process made on the basis of the adopted code word, after which a hard decision vector and a reliability vector are created. Bit reliability can be correlated with the communication channel. In one embodiment, bit reliability is created using known, calculated, or estimated properties of a communication channel, such as, but not limited to, the signal-to-noise ratio associated with signals propagating through the communication channel. In another embodiment, the set of bit reliability is initially correlated with the set of received bits, for example, but not only, with the absolute value of the signal received for each bit position. In one embodiment, one of a predetermined set of values may be selected as the threshold value, and this choice depends on the structure of the given code, the SNR of the channel, and the decoding iteration number.

In step 130, the bit reliability created based on the received codeword is updated based on a comparison with a threshold value. In various embodiments, a set of bit reliability can be created, compared with a threshold and updated. In step 140, the codeword is reconstructed based on comparisons of the calculated bit probabilities with a threshold where the calculated bit reliability is obtained from the originally created bit reliability. If the recovered codeword does not pass the validation test, the process continues with another iteration. The reconstructed codeword can be created by inverting one or more bits of the codeword version in a given iteration. At step 150, the reconstructed codeword is provided at the end of the iterative decoding process. If the decoding process is successful, then it is considered that the reconstructed codeword is a codeword that was provided at the transmission site at the other end of the communication channel. If a valid codeword is not provided during iterative decoding at the end of several iterations, it is advisable to retransmit the codeword. In one embodiment, this method is used to decode the LDPC codeword.

2 shows an embodiment of an apparatus 200 for iteratively decoding information received from a communication channel. The device 200 includes a receiver 210, a bit reliability generator 220, and a comparison unit 230. A receiver 210 receives data from a communication channel. In one embodiment, the received data is presented as bits in the signal stream, where these bits define a codeword. Alternatively, receiver 210 may receive a signal and convert the received signal to a code word. A bit reliability generator 220 generates one or more bit reliability associated with a received codeword. In one embodiment, the initial bit reliability is generated based on the received codeword and the likelihood ratio related to the probabilities that the binary signal is one or zero. These bit reliability can also be correlated with the signal-to-noise ratio of the communication channel from which the received codeword is obtained. Alternatively, the initial bit reliability may be generated based on the received codeword, for example, by setting the initial bit reliability to the absolute value of the received codeword for each bit.

A comparison unit 230 compares the bit reliability generated by the bit reliability generator with a threshold. After performing the specified comparison, the comparison unit 230 updates the bit reliability and bits of the test codeword depending on the results of the comparison with the threshold value. The comparison unit 230 uses a plurality of threshold values depending on the current iteration in the decoding process. The updated bits of the test codeword are evaluated to determine if the updated bits provide a valid codeword. If a valid codeword is generated, then the comparison unit 230 may issue a valid codeword or extract information bits (message bits) from the codeword and output a message. Block 230 comparison can be implemented in the form of a hardware configuration or made in the form of a software configuration. In one embodiment, the threshold values used by the comparison unit 230 are set in accordance with the iteration in which the value is used. You can use the threshold table that associates a predetermined value with the iteration number specified for a given code structure and a given situation in the channel, for example, respectively, of the SNR of the channel. In another embodiment, the threshold values used by the comparison unit 230 are set using bit reliability and the comparison result with the threshold value set for the decoding iteration in which the comparison unit 230 is currently operating. In one embodiment, device 200 decodes the LDPC codeword.

Figure 3 shows a block diagram of a variant of the method for encoding information with the aim of transmitting a codeword representing this information over the communication channel. At step 310, information is received at the encoder as information bits. In step 320, information bits are encoded to provide a codeword. The encoding is implemented in such a way that the version of the encoded codeword after it is received is iteratively decoded using bit reliability, updated by comparison with the threshold, where the threshold is updated to provide multiple values during the iterative process. If the codeword version is correctly received at the other end of the communication channel through which the codeword was transmitted, then iterative decoding can provide the codeword without correcting the received version of the codeword. If the received version of the transmitted codeword contains errors, then the iterative decoding process provides a correction process using a plurality of thresholds. At step 330, the code word is transmitted over the communication channel. In one embodiment, this method is used to encode the LDPC codeword.

FIG. 4 shows an embodiment of a device 400 for encoding information that is iteratively decoded using a plurality of threshold values at a receiving end of a communication channel. The device 400 includes a receiver 410 for receiving information for encoding, an encoder 420 and a transmitter 430. A receiver 410 receives data provided to the device 400 for transmission over a communication channel. Encoder 420 uses this information and creates an encoded word containing information bits (message bits) and bits for error correction at the receiving end of the transmission channel. Information supplied to encoder 420 may be represented as information bits. Alternatively, encoder 420 may use the provided information and generate information bits. Encoder 420 encodes the information by creating a codeword, so that the version of this codeword, upon receipt, is iteratively decoded using bit reliability updated by comparison with a threshold. The threshold is updated to provide multiple values during the iterative process. A transmitter 430 sends a coded codeword over a communication channel.

A communication channel between a device that uses an embodiment of a decoding method using multiple thresholds and a device where an embodiment of an encoding method associated with said decoding method is used may include, but is not limited to: a fiber optic channel, a wired channel, a channel, reading data stored in a memory unit or a wireless channel. Additionally, embodiments of the invention for decoding and encoding can be adapted for use with any protocol for transmission over any communication channel.

5 is a block diagram of an embodiment of a communication network 500 comprising a system 502 for encoding a high-redundancy codeword with parity and a system 520 for decoding a version of a coded high-excess codeword with parity obtained from communication channel 505 between the two systems. The information source provides LDPC information bits to encoder 510 of system 502. The LDPC encoder 510 includes an encoding procedure 512 that uses an H matrix structure 514 to generate a codeword transmitted to channel 505. The H matrix structure 514 is designed using code parameters 514 LDPCs that are provided to system 502 and system 520.

System 520 includes an LDPC decoder 522 that uses LDPC code parameters 514. The LDPC code parameters are used by the H matrix structure 524 in the LDPC decoder 522. The codeword received from channel 505 is supplied to initial decision block 526 to obtain soft decisions, resulting in hard decisions 528 and reliability decisions 530. The hard decisions 528 are the initial binary bits for the test codeword, and the reliability decisions are the initial bit reliability associated with the bits of the test codeword. In one embodiment, hard decisions 528 are binary bits set in accordance with the sign of the corresponding bit of the received codeword. A received bit having a positive value is set to one, and a received bit having a zero value or a negative value is set to zero. Alternatively, a received bit having a zero value or a positive value is set to zero, and a received bit having a negative value is set to one. The initial bit reliability for each bit of the received codeword may be set equal to the absolute value of the corresponding bit of the received codeword.

Hard decisions 528, reliability decisions 530, and the H matrix structure 524 are provided to a multi-threshold decoder 532 to perform an iterative decoding process to provide information bits 508 from the system 520. Multi-threshold decoder 532 uses thresholds 534 with which the calculated reliability is compared and the bit reliability and the bits for the test codeword are updated. Design reliability is derived from bit reliability. In one embodiment, the calculated reliability is derived from initial reliability decisions and calculated checksums. Bits for the test codeword are created from hard solutions 528 with bits whose values are reversed, in accordance with the results of comparison with the threshold value at each iteration. In one embodiment, thresholds 534 are calculated based on a comparison of the calculated bit reliability and the threshold value for a given iteration. In one embodiment, a set of thresholds 534 is calculated based on a comparison of the calculated bit reliability and the threshold value for a given iteration, and these thresholds are used in several other iterations. The second set of thresholds 534 is set with a fixed value, such as zero, and is used in several other iterations. The fixed value for the threshold is used in a group of iterations that form the last iterations in the decoding process. Alternatively, thresholds 354 are set equal to predetermined values depending on the iteration in which this threshold value is used.

Various options may be implemented for communication network 500, system 502, and system 520. Communication network 500 may be a terrestrial network using optical fibers, coaxial cable, or wires as the transmission medium for communication channels between different nodes of communication network 500. The communication network 500 may be implemented as a wireless network in which systems 502 and 520 are configured as part of a node of a communication network 500 or as part of a wireless receiver that receives signals from nodes of a communication network 500. Systems 502 and 520 can be configured as part, but not only, of network components, such as base stations, and can be configured as part, but not only, of wireless receiving devices, such as portable communication devices, computers, and laptop computing devices ". Although not shown, systems 502 and 520 configured in the wireless communications system 500 include antennas for transmitting and receiving signals that contain one or more codewords. Decoding and coding schemes and devices do not limit the type or configuration of support components such as antennas in systems 502 and 520. In one embodiment, a substantially omnidirectional antenna is used.

Each node of the communication network 500, configured as a terrestrial network or as a wireless network, may contain a system version 502 and / or a system version 520, or a system that is a combination of a system 502 and a system 520. Additionally, the decoding and encoding processes used in the encoders and decoders of systems 502 and 520, as well as other options described here for processes and a decoding and encoding device, can be adapted to communication standards and protocols used in communication network 500.

FIG. 6 is a flowchart of an embodiment of a method that can be used with an LDPC decoder, such as LDPC decoder 522, with the configuration shown in FIG. 5. At step 610, the initial hard decision vector and the initial reliability vector are constructed. For the received codeword Y = (Y ₁ , Y ₂ , ..., Y _n ), let N (m) be the set of codeword positions that are used in the mth parity check: N (m) = {n: H _{m, n} = 1}, where H is the parity check matrix, such as that provided by the H matrix structure 524 in FIG. 5, and H _{m, n} = 1 means that the value of the matrix H at position m, n is equal to one. In addition, we assume that M (n) is a set of parity checks that includes the nth position of the codeword: M (n) = {m: H _{m, n} = 1}.

битовая надежность контроля четности определяется как Y_mn, где первоначально Y_mn = R_n, а соответствующее жесткое решение определяется как X_mn, где первоначально X_mn = X_n. Upon initialization, for each element y _{i of the} received vector, such as that received in the LDPC decoder 522 of FIG. 5, the hard decision X _i and the bit reliability R _i are calculated. The bit reliability R _{i is} set equal to the absolute value of Y _i . A tough decision can set X _i equal to one if Y _{i is} greater than zero, and X _i equal to zero if Y _{i is} less than or equal to zero. For everybody

the parity check bit reliability is defined as Y _mn , where initially Y _mn = R _n , and the corresponding hard decision is defined as X _mn , where initially X _mn = X _n.

контрольные суммы S_mn вычисляются как:After initialization, at step 620, the iterative process begins by performing a calculation of the reliability of the parity. In one embodiment, for each n and each

checksums S _mn are calculated as:

Σ X_mn' mod 2, где суммирование выполняется по

N(m)/n. Выражение N(m)/n означает набор N(m), где исключен бит n. Значение минимума Y_mn определяется как:

Σ X _{mn '} mod 2, where the summation is performed over

N (m) / n. The expression N (m) / n means the set N (m), where bit n is excluded. The minimum value of Y _mn is defined as:

N(m)\n.Y _{mn min} = min {Y _{mn '} } for

N (m) \ n.

At step 630, a threshold is calculated. In one embodiment, the threshold is set equal to the minimum value for all n from the set of updated bit reliability defined for each n as

.Z _n = R _n + Σ (-1) ^Smn Y _{mn min} for

.

These updated bit reliability will also serve as reliability for comparison in the decoding process. In one embodiment, a multi-valued threshold is provided by calculating a value for the threshold during each iteration of the decoding process. This approach is adaptive. In one embodiment, a multi-valued threshold can be provided by calculating the value for the threshold during the first iteration using updated bit reliabilities and provided as a fixed value, such as zero, for the remaining iterations. In one embodiment, a multi-valued threshold can be provided by calculating the value for the threshold during the first iteration using updated bit reliabilities and using this value for the first half of the total number of iterations and then using a fixed value, for example, zero, for the other half of all iterations.

In one embodiment, a multi-valued threshold can be provided by using a predetermined set of thresholds where the value used is correlated with the iteration number in which it is applied. In one embodiment, predetermined values are assigned in accordance with the parameters of the communication channel from which the received codeword is derived. These parameters may include a set of signal-to-noise ratios, a set of other channel parameters, or a set of standard (mean square) deviations correlated with channel parameters. In one embodiment, a predetermined set of parameters can be determined through a modeling process. One such process may include modeling a decoding process to provide a set of thresholds for a given signal-to-noise ratio and setting a threshold for a given signal-to-noise ratio as an average of a set of simulated thresholds. This process can be performed for each signal to noise ratio in the set of signal to noise ratios. You can implement other statistical variations in the simulated thresholds for the decoding process. These predefined values, whichever option is used to provide these values, can be stored in a table. In addition to the software implementation, these tables facilitate hardware implementations by using a quick reference table.

In one embodiment, thresholds are selected for a plurality of decoding operating modes. In one mode, one or two errors are corrected during decoding. In this mode, the decoding process does not introduce an additional error into the codeword vector. In another embodiment, the decoding process corrects more than one error in one iteration and can correct one or two errors in a subsequent iteration.

Various combinations can be used to provide many threshold values during the iteration process, depending on the permissible complexity for a given application.

величина Z_mn определяется какAt step 640, the parity check bit reliability is updated. A set of reliability is calculated for comparison based on the reliability of the parity, which can be considered as a vote for parity. For each n and each

the value of Z _mn is defined as

N(n)/m.Z _mn = R _n + Σ (-1) ^Sm'n Y _{m'n min} for

N (n) / m.

For each n, the parity bit reliability is updated as the corresponding hard decisions X _mn based on the reliability for comparison of Z _mn . A reliability update or a tough decision includes maintaining a reliability or tough decision if the comparison shows that the reliability or tough decision should remain the same.

{Z _mn, Z _mn > 0 Y _mn = {-Z _mn, Z _mn <threshold {0; otherwise X _mn Z _mn > threshold X _mn = { 1-X _mn, Z _mn <threshold

In step 650, a new vector C of a robust reliability solution is created for comparison or updated bit reliability Z _n in the form

X _i Z _i > 0 C _i = { 1-X _i Z _i <0.

Updating a bit involves storing the bit value if the comparison shows that the bit should remain the same. Members C _i provide bits of a test codeword to determine whether a valid codeword has been generated. At step 660, an assessment of the validity of the test codeword is performed. If the following relationship appears

CH ^T ≠ 0,

then the iterative process returns to step 620 using the updated parity reliability values.

Another embodiment of a method for iteratively decoding an LDPC codeword using a threshold having a plurality of values includes assigning to the threshold only two values during an iterative process. The first value is set equal to the minimum of the updated reliability Z _n for Z _n created in the first iteration before making comparisons of solutions. The second value is set to zero for subsequent iterations. This option provides a gain of 0.5 dB during decoding compared with other known fast decoding schemes having virtually the same complexity.

In embodiments for multi-threshold schemes, any LDPC code is decoded in any situation. However, for a predetermined code and known channel properties, thresholds can be selected more accurately. In various embodiments, the number of iterations can be taken equal to log ₂ n, where n is the length of the code word. This provides the maximum number of iterations to be performed until the completion of the iteration before the restoration of a valid codeword. Experiments show that tuning and optimizing the threshold provides an increase in BER gain, but determining the exact relationship between the thresholds, code structure, and channel properties depends on the particular application.

Various variations or combinations of variations for a device and methods for decoding and encoding communication signals using the multi-threshold schemes described herein may be implemented in hardware, software, and combinations of hardware and software. These implementations may include computer-readable media having computer-readable instructions for performing a decoding and / or encoding method using an embodiment of a scheme where multiple thresholds are used for iterative encoding. Machine-readable media is not limited to any one type of media. The computer-readable medium used depends on the application using the multi-threshold scheme option. In one embodiment, a processor connected to a receiver and / or transmitter may use various types of memory to which it is connected and a computer-readable medium for controlling decoding and / or encoding of information for data transmission. In one embodiment, the parameters used to iteratively decode the received codeword may be stored in memory.

Although a multi-valued threshold method variant can be used with any LDPC code, one example is provided below, which is not a limitation. This example represents an embodiment of a decoding method using a multi-threshold decoder for a high redundancy parity check code based on Reed-Solomon codes with two information symbols (RS-LDPC code). This is an example (n, γ, ρ) of the RS-LDPC code with n = 32, γ = 4, and ρ = 3, and the speed R is approximately equal to 0.4, and the parity matrix is as follows:

H _RS-LDPC =

10,000,000 10,000,000 10,000,000 10,000,000 00001000 00000010 00000100 01000000 00000100 00000001 00000010 00100000 00000010 01000000 00000001 00010000 00000001 00100000 01000000 00001000 01000000 00010000 00100000 00000100 00100000 00001000 00010000 00000010 00010000 00000100 00001000 00000001 10,000,000 00001000 00000001 01000000 00001000 00000001 01000000 10,000,000 00000100 00000010 00001000 00000010 00000010 00010000 10,000,000 00001000 00000001 00000100 00000100 00010000 01000000 01000000 00010000 00000001 00100000 10,000,000 00100000 00100000 00010000 00100000 00000010 00000100 10,000,000 00000100 01000000 00100000 00001000 00010000 00000001 00000010 00000100 01000000 00100000 10,000,000 00000010 00000001 00000100 00000001 00000001 00001000 10,000,000 00000100 01000000 00000010 00000010 00001000 00100000 00100000 00001000 01000000 00010000 10,000,000 00010000 00010000

For this example, the received vector Y = {Y _i } has a codeword length of 32:

+0.22 +1.47 +0.52 +0.83 -2.38 +0.71 +1.44 -1.95 +0.08 -0.14 +1.19 +0.19 -1.86 -0.16 +0.94 +0.05 +1.01 -0.42 +0.78 -1.54 -0.84 +1.44 +1.65 -0.45 +0.94 -2.08 -1.31 +1.91 -0.69 +1.04 +1.39 +0.83

In this embodiment, the number of iterations is set to log ₂ n, where n is the codeword length. For n = 32, the number of iterations is 5. Thresholds for iterations 1, 2, and 3 are set according to the minimum of updated bit reliability Z _n calculated in the first iteration. Thresholds for iterations 4 and 5 are set to 0.

When initializing the hard decision vector, X = {X _i } as follows:

1 1 1 1 0 1 1 0 1 0 1 1 0 0 1 1

1 0 1 0 0 1 1 0 1 0 0 1 0 1 1 1

bit reliability R _i are

+0.22 +1.47 +0.52 +0.83 +2.38 +0.71 +1.44 +1.95 +0.08 +0.14 +1.19 +0.19 +1.86 +0.16 +0.94 +0.05 +1.01 +0.42 +0.78 +1.54 +0.84 +1.44 +1.65 +0.45 +0.94 +2.08 +1.31 +1.91 +0.69 +1.04 +1.39 +0.83

and the reliability of the parity is initially set as

Y _mn :

At the first iteration, checksums are calculated, and the minimum reliability of the parity control Y _mn-min is defined as

checksums S _mn :

Y _mn-min :

The threshold is calculated as a result of the first calculation of the updated bit reliability (bit reliability for comparison) Z _n as

Z _n :

From these values of Z _{n it} is determined that the threshold is -1.04.

Reliability is formed for comparison Z _mn , and the reliability of the parity check Y _mn and the corresponding hard decisions X _mn are updated as follows

Z _mn :

Y _mn :

X _mn :

Then, a test codeword is generated in the form of a reconstructed codeword

C = {C _i }:

0 1 1 1 0 0 1 0 0 0 1 1 0 0 1 1

1 1 1 0 0 1 1 0 1 0 0 1 1 1 1 1

The test codeword C is such that HC ^T ≠ 0, and the second iteration is performed. Iteration 2 begins with the calculation of new checksums and new minimum reliability parity control Y _mn-min as checksums S _mn :

Y _mn-min :

The threshold for iterations 1, 2, and 3 is assigned the minimum value of the updated bit reliability (bit reliability for comparison) Z _n calculated in the first iteration. Thus, in this iteration, the threshold is set to -1.04, as in the first iteration. Updated bit reliability (bit reliability for comparison) Z _n for the second iteration are defined as:

Z _n :

Reliability is formed for comparison Z _mn for the second iteration, and parity reliability Y _mn for the second iteration, and the corresponding hard solutions X _mn are updated as follows.

Z _mn :

Y _mn :

X _mn :

Then, a test codeword is generated in the form of a reconstructed codeword

C = {C _i }:

0 1 1 1 0 0 1 0 0 0 1 1 0 0 1 1

1 0 1 0 0 1 1 0 1 0 0 1 1 1 1 1

The test codeword C is such that HC ^T = 0, and decoding ends because it is considered successful. In this example, iterations 3-5 are not required, and the iterations are completed. Although the second value was chosen, which was planned to be used in iterations 4 and 5, the need for it disappeared. This example implementation of a multi-threshold decoding scheme should not be construed as limiting for various embodiments of the present invention.

Various options for decoding a received codeword using a multi-valued threshold can be applied to any high-redundancy parity code. In these embodiments, when the weight of the parity check decreases, the decoding quality increases. These options can be applied in a method that does not greatly complicate the decoding scheme. Depending on the permissible complexity for a given application, various combinations can be used to provide multiple threshold values during an iterative process. The use of variants of the multi-threshold decoding method provides an additional reduction in complexity, since these variants can operate in hardware implementations without floating-point numbers, which simplifies and reduces the cost of the overall technical solution. Variants of a multi-threshold decoding scheme allow pipelining and / or parallel processing in hardware implementations, since the use of a multi-valued threshold can provide a fast decoding scheme.

Variants of a multi-threshold decoding scheme can be adapted for use in any system operating with a communication channel. The communication channel may be part of a terrestrial communications network or a wireless communications network. Embodiments of the present invention can indeed be successfully implemented as part of any wireless system where multi-carrier wireless communication channels are used (eg, orthogonal frequency division multiplexing (OFDM), discrete multi-tone modulation (DMT), etc.), for example can be used without limitation in wireless personal area network (WPAN), wireless local area network (WLAN), wireless urban network WMAN), wireless wide area network (WWAN), cellular network, third generation network (3G), network fourth generation (4G), universal mobile telephone system (UMTS) and similar communication systems.

Although specific embodiments of the invention have been shown and described above, it should be apparent to those skilled in the art that the specific embodiments presented can be replaced by any structure that is calculated to achieve the same goal. It is intended that this application cover any adaptations or modifications to the variations of the present invention. It should be clear that the above description is intended to illustrate and not limit, and that the phraseology or terminology used here is adopted for the purpose of description and not limitation. Based on the foregoing description, those skilled in the art will readily be able to propose combinations of the above options and other embodiments of the invention. The scope of the present invention includes any other applications that use variants of the above structures and implementation methods. The scope of embodiments of the present invention should be determined according to the claims, together with the full scope of its equivalents.

Claims (46)

1. A method for iteratively decoding a received codeword, comprising receiving a codeword from a communication channel, the codeword being a highly redundant parity-controlled codeword; performing a soft decision process for each bit of the received codeword to generate a received vector from which the hard decision vector and initial bit reliability are generated; iteratively decoding the received codeword using bit reliability, so that each iteration includes selecting a value for the threshold, the threshold being updated to provide a plurality of threshold values during iterative decoding, so that at least two different threshold values are used to iteratively decode the received codeword, the initial value for the threshold being calculated using the initial bit reliability; updating each bit reliability based on a comparison using a threshold value for iteration; and restoration of the received codeword based on a comparison of the updated bit reliability with a threshold. 2. The method of claim 1, wherein iteratively decoding the received codeword includes using a non-zero value for a threshold in the first iteration and using zero as a threshold value in other iterations. 3. The method of claim 1, wherein the method includes correcting one or more bits of a test codeword based on comparing the updated bit reliability with a threshold to provide a valid codeword, wherein the test codeword is initially generated from the received codeword. 4. The method according to claim 1, in which the use of a threshold includes using a threshold with a value calculated using the minimum reliability for comparison from a set of reliability for comparison, each reliability for comparison is calculated based on the initial bit reliability and a set of parity reliability. 5. The method according to claim 1, in which the generation of initial bit reliability includes the generation of initial bit reliability correlated with the communication channel. 6. The method according to claim 1, in which for each bit of the received vector bit reliability is generated in the form of the absolute value of the bits of the received vector. 7. The method according to claim 1, in which the method includes calculating the reliability of the parity for each parity from a set of parity checks of the received codeword. 8. The method according to claim 7, including the formation of the first reliability for comparison based on the initial bit reliability related to the bit of the hard decision vector, and on the basis of the first subset of the reliability of the parity; comparison of the first reliability for comparison with a threshold; updating the reliability of parity based on a comparison of the first reliability for comparison with a threshold; generating a second reliability for comparison based on the initial bit reliability, and on the basis of a second subset of the parity reliability; updating the bit of the test codeword based on the comparison of the second reliability for comparison with the set value; and determining whether the test codeword is a valid codeword. 9. The method of claim 8, wherein updating the bit of the test codeword includes reversing the value of the bit. 10. The method of claim 8, in which updating the bit of the test codeword based on the comparison of the second reliability for comparison with the set value includes updating the bit of the test codeword based on the comparison with zero of the second reliability for comparison. 11. The method according to claim 1, in which iterative decoding of the received codeword includes performing several iterations, the number of iterations being equal to log ₂ n, where n is the length of the codeword code. 12. A method for encoding information bits of a codeword, comprising receiving information bits in the encoder; providing parameters of a high-redundancy code with parity, which are also provided to the decoding process; encoding information bits to provide a high-redundancy codeword with parity using parameters of a high-redundancy codeword with parity, so that the received version of the encoded codeword is iteratively decoded using bit reliability updated by comparison with a threshold where the threshold is updated to provide many values during the iterative process; and transmitting a highly redundant code word with parity through a communication channel. 13. The method of claim 12, wherein transmitting the highly redundant parity codeword over a communication channel includes transmitting a highly redundant parity codeword over a fiber optic channel. 14. The method according to item 12, in which the transmission of a highly redundant code word with parity through a communication channel includes transmitting a highly redundant code word with a parity through a channel that reads data stored in the memory unit. 15. The method according to item 12, in which the transmission of the code word includes a wireless transmission of the code word. 16. A computer-readable medium having computer-executable instructions for executing a method comprising receiving a codeword from a communication channel, wherein the codeword is a highly redundant codeword with parity; performing a soft decision process for each bit of the received codeword to generate a received vector from which the hard decision vector and initial bit reliability are generated; iteratively decoding the received codeword using bit reliability, so that each iteration includes selecting a value for the threshold, the threshold being updated to provide a plurality of threshold values during iterative decoding, so that at least two different threshold values are used to iteratively decode the received codeword, the initial value for the threshold being calculated using the initial bit reliability; updating each bit reliability based on a comparison using a threshold value for iteration; and restoration of the received codeword based on a comparison of the updated bit reliability with a threshold. 17. The computer-readable medium of claim 16, wherein the method includes correcting one or more bits of a test codeword originally created from the received codeword to provide a valid codeword. 18. The computer-readable medium of claim 16, wherein the method includes providing a message word from a received codeword after iteratively decoding the received codeword. 19. The computer readable medium of claim 16, wherein generating the initial bit reliability includes generating initial bit reliability correlated with the communication channel. 20. The computer-readable medium of claim 16, wherein using a threshold includes using a threshold with a value calculated using the minimum reliability for comparison from a set of reliability for comparison, and each reliability for comparison is calculated on the basis of the initial bit reliability and the set of reliability of the parity. 21. The computer-readable medium of claim 16, wherein the computer-readable medium is a component of a wireless device. 22. Machine-readable medium having computer-executable instructions for executing a method comprising receiving information bits in the encoder; providing parameters of a high-redundancy code with parity, which are also provided to the decoding process; encoding information bits to provide a highly redundant parity-controlled codeword using the parameters of a high-redundancy parity code, so that the received version of the encoded codeword is iteratively decoded using bit reliability updated by comparison with a threshold where the threshold is updated to provide a plurality of values in the course of the iterative process; and transmitting a highly redundant code word with parity through a communication channel. 23. The machine-readable medium of claim 22, wherein transmitting the highly redundant parity codeword through a communication channel includes transmitting a highly redundant parity codeword through a fiber optic channel. 24. The machine-readable medium of claim 22, wherein transmitting the highly redundant parity codeword through a communication channel includes transmitting a highly redundant parity codeword through a channel reading data stored in a memory unit. 25. The computer-readable medium of claim 22, wherein the computer-readable medium is a component of a wireless system. 26. An apparatus for iteratively decoding a received codeword, comprising: a receiver for receiving a codeword from a communication channel, wherein the codeword is a highly redundant codeword with parity; a bit reliability generator for providing a set of bit reliability, each bit reliability corresponding to a bit of a received codeword received by the receiver; a soft decision unit for processing each bit of the received codeword to generate a received vector from which the hard decision vector and the set of bit reliability are generated; and a comparison unit configured to use a threshold, so that for each iteration of the iterative process to decode the received codeword, each bit reliability is updated based on a comparison using the threshold value selected for iteration, where the threshold is updated to provide multiple threshold values during iterative decoding, so that at least two different threshold values are used to iteratively decode the received codeword, and the initial value for and is calculated using the initial bit reliabilities. 27. The device according to p, in which the generator of initial bit reliability is configured to generate initial bit reliability correlated with the communication channel. 28. The device according to item 27, further comprising a first reliability calculation unit for calculating a parity check reliability for each parity from a set of parity checks of a received codeword; a second reliability calculation unit for calculating a first reliability for comparison based on the initial bit reliability related to the bit of the hard decision vector and the first subset of the parity reliability; a first comparison unit for comparing the first reliability for comparison with a threshold for updating the reliability of the parity based on comparing the first reliability for comparison with a threshold; a third reliability calculation unit for calculating a second reliability for comparison based on an initial bit reliability and a second subset of parity reliability; a second comparison unit for determining whether the test codeword is a valid codeword, where the test codeword is a version of the received codeword updated based on a second reliability comparison for comparison with a set value. 29. The device according to p, in which each unit of calculation of reliability and each unit of comparison are included in a common threshold decoder. 30. The device according to clause 29, in which the common threshold decoder is adapted to provide information bits from a valid code word. 31. The device according to p, in which the device is a wireless device. 32. A device for encoding information bits of a codeword, containing transmitter; a high redundancy parity encoder for encoding information bits into a high redundancy parity code word for transmission over a communication channel by a transmitter, so that the version of a high redundancy parity code word is received iteratively decoded using bit reliability updated by comparison with a threshold where the threshold is updated, to provide many meanings during an iterative process; and an h-matrix structure for using parameters of a high redundancy parity check code common to decoding units receiving a transmitted version of a high redundancy codeword with parity, the h-matrix structure being used by a high redundancy parity encoder to generate a high redundancy parity codeword. 33. The device according to p, in which the communication channel includes a fiber optic channel. 34. The device according to p, in which the communication channel includes a channel that reads the data stored in the memory unit. 35. The device according to p, and the device is a stationary wireless device. 36. A system for iteratively decoding a received codeword, comprising CPU; a substantially omnidirectional antenna for receiving a signal; a receiver for receiving the received codeword from the signal, the received codeword being a highly redundant parity-controlled codeword; a decoder for iteratively decoding the received codeword received by the receiver, the decoder including a bit reliability generator for providing a set of bit reliability, each bit reliability corresponding to a bit of a received codeword; a soft decision unit for processing each bit of the received codeword to generate a received vector from which the hard decision vector and the set of bit reliability are generated; and a comparison unit configured to use a threshold, so that for each iteration to iteratively decode the received codeword, each bit reliability is updated based on a comparison using the threshold value selected for iteration, where the threshold is updated to provide multiple threshold values during iterative decoding, so that at least two different threshold values are used to iteratively decode the received codeword, and the initial value for the threshold is calculated It is a C using the initial bit reliabilities; and a memory connected to the processor, the memory being intended for storing parameters used for iteratively decoding the received codeword. 37. The system of claim 36, wherein the system is a portable wireless system. 38. The system of claim 36, wherein zero is a threshold value in iterations other than the first iteration. 39. The system of claim 36, wherein the bit reliability generator and the comparison unit are components of a threshold decoder adapted to use updated threshold values to determine a valid codeword and provide information bits from a valid codeword. 40. A system for encoding information bits of a codeword, comprising CPU; transmitter; a high redundancy parity encoder for encoding information bits into a high redundancy parity code word for transmission over a communication channel by a transmitter, so that the version of a high redundancy parity code word is received iteratively decoded using bit reliability updated by comparison with a threshold where the threshold is updated, to provide many meanings during an iterative process; an h-matrix structure for using parameters of a high redundancy parity check code common to decoding blocks receiving a transmitted version of a high redundancy codeword with a parity check, wherein the h-matrix structure is used by a high redundancy parity encoder to generate a high redundancy codeword with a parity check; a memory connected to the processor, the memory being intended for storing parameters of a highly redundant code with parity; and essentially an omnidirectional antenna connected to a transmitter. 41. The system of claim 40, wherein the system is a stationary wireless system. 42. The system of claim 40, wherein the communication channel includes a fiber optic channel. 43. A system for iteratively decoding a received codeword, comprising CPU; a receiver for receiving a received codeword in a signal from a terrestrial communication channel, wherein the codeword is a highly redundant codeword with parity; a decoder for iteratively decoding the received codeword received by the receiver, the decoder including a bit reliability generator for providing a set of bit reliability, each bit reliability corresponding to a bit of a received codeword; a soft decision unit for processing each bit of the received codeword to generate a received vector from which the hard decision vector and the set of bit reliability are generated; and a comparison unit configured to use a threshold, so that for each iteration to iteratively decode the received codeword, each bit reliability is updated based on a comparison using the threshold value selected for iteration, where the threshold is updated to provide multiple threshold values during iterative decoding, so that at least two different threshold values are used to iteratively decode the received codeword, and the initial value for the threshold is calculated It is a C using the initial bit reliabilities; and a memory connected to the processor, the memory being intended for storing parameters used for iteratively decoding the received codeword. 44. The system according to item 43, and the system is adapted for connection to a fiber optic communication network. 45. The system of claim 43, wherein zero is a threshold value in iterations other than the first iteration. 46. The system of claim 43, wherein the bit reliability generator and the comparison unit are components of a threshold decoder adapted to use updated threshold values to determine a valid codeword and provide information bits from a valid codeword.

Images