RU2786023C1 - Method for message transmission in systems with feedback and hybrid automatic repeat request - Google Patents

Abstract

FIELD: telecommunications.

SUBSTANCE: invention relates to the field of telecommunications, namely to the transmission of digital information (data) using reverse channels to detect and prevent errors. To achieve the effect, if an error is detected, the code word on the receiving side is not erased, and the code word generated on request on the transmitting side contains both new, previously not transmitted information, and re-transmitted information, which makes it possible to correct errors in the previously stored word. Error detection is performed by encoding the message symbols with an error-correcting code on the transmitting side and decoding with error detection on the receiving side. The decision about the presence of errors or their absence is transmitted over the reverse channel, if errors are detected, the previously transmitted code word is repeated (the previously transmitted information is duplicated).

EFFECT: increasing the throughput of the data transmission system in channels with changing transmission conditions.

1 cl, 3 dwg

Description

The invention relates to the field of data transmission using decisive feedback in the protocols of computing and information systems. Methods of transmission with a decisive feedback implement a retransmission of a message encoded with an error-correcting code if errors are detected on the receiving side. The principles for constructing such systems (systems with automatic repetition request - ARQ, ARQ) are described, for example, in the monograph S.Lin and D.J.Costello, Jr. Error Control Coding: Fundamentals and Applications. NJ: Englewood Cliffs. 2001. The current state of the art is presented in the review article A. Ahmed, A. Al-Dweik, Y. Iraqi, H. Mukhtar, M. Naeem and E. Hossain, "Hybrid Automatic Repeat Request (HARQ) in Wireless Communications Systems and Standards: A Contemporary Survey, IEEE Communications Surveys & Tutorials, vol.23, no.4, pp.2711-2752, Fourthquarter 2021, doi: 10.1109/COMST.2021.3094401.

In traditional automatic repeat request (ARQ) methods, errors in transmitted packets are checked at the receiving end with error detection code (typically cyclic redundancy check (CRC) or other suitable redundancy code). If no errors are detected in the frame during the control at the receiving side, the receiving side sends a notification of successful receipt (ACK) to the recipient. If an error is detected in the frame, the receiving side sends a Negative Acknowledge Acknowledgment (NAK) signal, requesting a retransmission.

User data can be further protected with an error-correcting code that increases the likelihood of successful transmission. Such ARQ schemes that combine the principle of ARQ with partial error correction coding are known as hybrid ARQ (HARQ) methods.

A measure of the effectiveness of the ARQ method is the average transmission rate (throughput), defined as the average number of bits of user data received at the receiving end in the time it takes to transmit one bit, as well as the probability of an undetected error and the average delay time. The level of redundancy of the error correction code used in the HARQ scheme has two opposite effects on the efficiency of the method: with increasing redundancy, the probability of successful transmission increases, but the proportion of user data in the codeword decreases. Typically, a fixed rate code is chosen that satisfies the throughput requirements for fixed channel characteristics.

In real channels with changing transmission conditions (changing the signal-to-noise ratio (SNR)), such as mobile and satellite channels, HARQ-Type 2 methods are used with the accumulation and processing of received repeated combinations (HARQ-CC) or with the transmission of additional redundant characters (with incremental redundancy, HARQ-IR). Such methods exhibit higher throughput by adapting error correction code redundancy to different channel conditions.

Description of the prior art

Known analogue of the invention, relating to methods with the accumulation of received combinations (HARQ-CC) METHOD OF TRANSMISSION-RECEPTION OF MESSAGES IN COMMUNICATION SYSTEMS. RF patent for IZ No. 2595627, 2015 (Analogue 1). The technical result of the invention is to increase the speed of receiving information by reducing the likelihood of re-questions.

по модулю 2, где l≠t, определяют число стертых кадров в каждом из (j+1) столбцов, при стирании одного кадра m_s,t восстанавливают его путем замены на

по модулю 2 где v≠s, если в полученном двумерном массиве i×j нет стертых кадров, к информационных бит каждого из h кадров выдают получателю, иначе повторно восстанавливают стертые кадры в каждых из i+1 строк и j+1 столбцов, если и после повторного просмотра остались стертые кадры, по каналу обратной связи производят переспрос этих h кадров.The method of transmitting and receiving messages in communication systems consists in the fact that on the transmitting side the original message is divided into portions of k bits, frames are formed from them with the HDLC protocol structure using the ARQ method for h steps back, h frames are transmitted over the communication channel, where h >3, on the receiving side, using the control sequence of the HDLC protocol, erroneous frames are detected, if there are no erroneous frames, information k bits are given to the recipient or the frame with errors is erased, the reception of subsequent h-1 frames is blocked, a request command is sent over the reverse channel and the transmitting side retransmits these h frames, characterized in that h frames are pre-arranged in a two-dimensional array of i rows and j columns h=i×j for each s-th row, where s = 1, 2 ... i, the first set of control frames m _{s is calculated, j+1} , equal to the sum of row frames modulo two, calculate the second set of control frames m _i+1,t equal to the sum of column frames modulo two, where t = 1, 2 ... J, the generated array and i+1 rows and j+1 columns are transmitted to the communication channel, after receiving in the received array, frames with errors are detected by the control sequence of the HDLC protocol, they are erased, the number of erased frames in each of i+1 rows is determined, if one frame m is erased _s,t restore it by replacing it with modulo 2, where l≠t, determine the number of erased frames in each of the (j+1) columns, when erasing one frame m _s,t restore it by replacing: with

modulo 2, where l≠t, determine the number of erased frames in each of the (j+1) columns, when erasing one frame m _s,t restore it by replacing it with

modulo 2 where v≠s, if there are no erased frames in the received two-dimensional array i×j, k information bits of each of h frames are given to the recipient, otherwise the erased frames are re-restored in each of i+1 rows and j+1 columns, if and after re-viewing, erased frames remain, these h frames are re-quested via the feedback channel.

In this method, when a detected error occurs, the information transmitted on request is duplicated. In addition, retransmitted frames are encoded with an iterative code, additional redundant information is transmitted to the channel, which is used to correct errors on the receiving side.

The disadvantage of the invention under consideration is the complete duplication of the block transmitted on request, the limited corrective ability of iterative coding, which reduces the transmission rate. According to the author, the method has a limited scope of application: the use of a protocol with blocking and returning to h steps (similar to HDLC) and is suitable for channels with a large signal propagation delay.

Known analogue related to the methods of HARQ IR. RF patent for IZ No. 2239952 METHOD OF MESSAGE TRANSMISSION IN SYSTEMS WITH FEEDBACK (analogue 2).

The technical result of the proposed method is to increase the speed of information transfer in the communication channel and simplify the software and hardware implementation of the method.

The method of transmitting messages in a feedback system consists in that, on the transmitting side of the communication system, an array of input information is divided into blocks of a certain length, which are encoded with an error-correcting code and then transmitted to the communication channel, on the receiving side of the communication system, the error-correcting code is decoded with detection and error correction, when errors are detected in the error-correcting code, the multiplicity of which lies outside the correcting capacity of the error-correcting code, a request is formed for retransmission of the error-correcting code, this request is transmitted via the feedback channel to the transmitting side and, upon request, additional symbols of the error-correcting code are transmitted, then when additional symbols of the error-correcting code are received, this code is re-decoded, characterized in that a shortened error-correcting cascade code is transmitted to the communication channel, on the receiving side of the communication system based on the results of decoding the inner code of the shortened n of the error-correcting concatenated code, first the number of erased words of the inner code is determined, then the number of transformed words of the inner code is estimated, then the number of additional words of the inner code of the error-correcting concatenated code is determined, necessary for the correct decoding of the outer code of the error-correcting concatenated code, and a request is formed to transmit these additional words of the inner code of the error-correcting concatenation code cascade code.

In the analog under consideration, additional redundant information is transmitted upon request, representing the words of the internal code of the cascading code, which reduces the throughput of the transmission system and increases the delay. The disadvantages of the method include limited applicability in existing protocols for the exchange of computer networks due to the use of only concatenated codes and the need to report to the transmitting side an estimate of the number of additional words of the internal code.

Closest to the proposed method is the method described in M. S. Mohammadi, I. B. Collings and Q. Zhang, "Simple Hybrid ARQ Schemes Based on Systematic Polar Codes for IoT Applications," in IEEE Communications Letters, vol. 21, no. 5, pp. 975-978, May 2017, doi: 10.1109/LCOMM.2017.2662012.

The technical result of the proposed method is to increase the speed of information transmission in the communication channel by reducing the likelihood of repeated request.

The transmission method uses a HARQ-IR scheme based on systematic polar codes or other code structures that allow redundant bits to be divided into non-overlapping segments. A systematic polar code of arbitrary rate R = K / N, where K is the number of information symbols, N is the length of the code, allows you to divide the redundant bits into an arbitrary number of non-overlapping sets and sequentially transmit systematic bits, and then redundant segments, until the message is successfully decoded and delivered to the recipient.

The method proposed by the authors uses at the initial stage the encoding of a message with a length of (K - C) characters that are encoded by the error detection code C0, C is the number of check characters. An output vector of length K is denoted by Xs and is sent over the channel. Let us denote the set of transmitted systematic bits as T1. If the receiver detects an error in Xs, then on the transmitting side Xs is encoded with a systematic polar code C1 at a rate of R=K/N and a set of r1 redundant symbols P1 is sent, where r1 = |P1| is the total number of bits in P1. The polar code is then decoded using a procedure that includes all K systematic code bits plus the P1 bits, forming a vector T ₂ =T ₁ +r ₁ . The recovered vector Xs is decoded with error detection in the C0 code, and if errors are detected, the transmitter sends another disjoint subset of size r2 of the excess P2 bits. This procedure ends when the C0 checksum is consistent, but potentially continues until all redundant bits have been sent to the receiving side, i.e. when

The prototype has several disadvantages:

• does not provide speed gain in synchronous data transmission protocols due to the fact that portions of additional information, as a rule, have a length significantly less than the length of the transmitted packet, and transmission of additional information is not provided;

• implementation of the method requires a drive of considerable length and a significant delay of information;

• the method can be used for a limited class of codes.

Characteristics of information disclosing the essence of the invention

The previously known methods described here use the transmission of redundant information (check symbols), which are used to decode a previously stored codeword with a code whose redundancy increases due to the length with each repeat request. In this case, forward or forward error correction (forward error correction, FEC) is implemented. This way of using redundancy has the disadvantages described above.

It is proposed to obtain additional redundancy not by increasing the length of the code, but by decoding the stored code vector with a detected error using an error-correcting code of the same length, but at a lower rate. In this case, in the case of a repeat request, the code vector must be of the same length, contain information that allows decoding the stored vector in a code with high redundancy, as well as additional symbols that have not been transmitted before. Successful transmission of the second vector can then provide error recovery in the first (saved) vector, the recipient will then be given the information symbols contained in the two received vectors. By analogy with FEC, this error correction could be called reverse forward error correction.

The implementation of this approach is possible using linear error-correcting codes that can be represented as systems of nested codes, i.e. unions of subsets, each of which is a linear error-correcting code or its shift (adjacent class).

Such systems are formed by many types of codes - BCH, Reed-Solomon, Reed-Muller, polar and others. An exhaustive description of nested code systems is contained in the book by E.L. Bloch, V.V. Zyablov Linear concatenated codes, Moscow: Nauka, 1982, 229 p.

Let А ₀ [n,k ₀ ,d ₀ ]-linear code of length n with k ₀ information symbols, minimum distance d ₀ . For such a code, we will consider a system of nested codes and their cosets with parameters A _i [n,k _i ,d _i ], i=0,…,m. The following relations hold for the system:

где G₀ (k₀ х n) порождающая матрица кода A₀,G₁- (k₁ х n), порождающая матрица A₁, G₀₁- ((k₀-k1) х n) матрица, порождающая образующие смежных классов. Пусть код используется в режиме обнаружения ошибок, а код - в режиме с исправлением ошибок. В начале передачи от источника поступает k0 символов, кодируются с помощью умножения на матрицу, при этом вектор кода является смежным классом кода, номер которого определяется символами. Информационный вектор сохраняется и сохраняется номер смежного класса.Consider a system of two nested linear codes A ₀ [n,k ₀ ,d ₀ ] and A ₁ [n,k ₁ ,d ₁ ], such that A ₀ is the union of cosets A ₁ . Such a system is given by a generating matrix

where G ₀ (k ₀ x n) is the generating matrix of code A ₀ , G ₁ - (k ₁ x n) is the generating matrix of A ₁ , G ₀₁ is ((k ₀ -k1) x n) is the matrix generating generators of cosets. Let the code be used in error detection mode and the code in error correction mode. At the beginning of the transmission, k0 symbols arrive from the source, are encoded by matrix multiplication, with the code vector being a contiguous code class, the number of which is determined by the symbols. The information vector is stored and the coset number is stored.

If no errors occur, information symbols are given to the recipient and the transmitting side receives an ACK confirmation. The transmission process continues in this way further, until an error is detected in some code word.

In this case, the code word is not erased, the transmitting side receives a NACK request signal. The transmitter uses this signal to generate a new vector that contains a set of lengths of previously not transmitted symbols and a set that is the number of the adjacent code class to which the previous vector belonged.

декодер по номеру смежного класса определяет образующий смежного класса, переводит сохраненный вектор в смежный класс кода и декодирует с исправлением ошибок в коде А₁. При успешном декодировании получателю выдается k₀+k₁ символов. Заметим, что аналогичная ситуация в обычной системе с переспросом обеспечила бы выдачу только k0 символов. В случае прототипа передача 2n символов в канале также привела бы к выдаче только k0 символов.If no errors are detected in this received vector during code-detection decoding

the decoder, by the number of the adjacent class, determines the generator of the adjacent class, translates the stored vector into the adjacent code class, and decodes with error correction in the code A ₁ . Upon successful decoding, the recipient is given k ₀ +k ₁ symbols. Note that a similar situation in a conventional querying system would provide only k0 characters. In the case of the prototype, transmitting 2n symbols in a channel would also result in only k0 symbols being output.

The consideration can be supplemented with the following example. Let a system of two nested BCH codes A ₀ [31,25,4] and A ₁ [31,15,8] be used. Code A0 is used in the error detection mode, and code A1 in the mode of correcting three or less errors. Let errors of multiplicity T ≤ 3 occur in three successively transmitted blocks, and the fourth one is received without error. Then, in accordance with the description of the method, after receiving the fourth vector and correcting errors in the previous three, 25+15+15+15=70 binary symbols will be issued from the drive to the recipient. In the case of a regular ARQ, after three retries, the recipient will only receive 25 characters.

, где

- вероятность обнаружения ошибок при декодировании в коде А₀. Пример хорошо иллюстрирует возможный выигрыш по скорости, но для его реализации требуется приемный буфер значительной длины, что увеличивает среднюю задержку выдачи сообщения. Кроме того, использование процедуры исправления и обнаружения может увеличивать вероятность необнаруженной ошибки по сравнению со стандартным способом использования CRC16 или CRC32.For the example under consideration, one can obtain an obvious estimate of the average speed, which is valid for the case of a receiving accumulator for a codeword of infinite volume:

, where

- the probability of detecting errors during decoding in the code A ₀ . The example well illustrates the possible gain in speed, but its implementation requires a receive buffer of considerable length, which increases the average delay in issuing a message. In addition, the use of the correction and detection procedure may increase the probability of an undetected error compared to the standard way of using CRC16 or CRC32.

The proposed method is based on the use of a system of three nested codes, which makes it possible to achieve the best gain in speed and average message delay in real transmission conditions.

We use the system

имеют вид:

при этом размерности матриц

Выполняются очевидные условия

The corresponding generating matrices

look like:

while the dimensions of the matrices

Obvious conditions are met

The system of nested codes can be constructed in such a way that

We give a description of the main idea of the method.

символов

кодируются в коде

и сохраняются в накопителе

и выдается в канал. На приемной стороне выполняется проверка на наличие ошибок. Если в принятом векторе ошибок не обнаружено, в канал обратной связи поступает сигнал подтверждения ACK. После получения сигнала ACK на приемной стороне формируется и выдается в канал блок по такому же правилу.At the beginning of the transmission from the source to the encoder on the receiving side,

symbols

encoded in code

and stored in storage

and issued to the channel. The receiving side checks for errors. If no errors are found in the received vector, an acknowledgment signal ACK is sent to the feedback channel. After receiving the ACK signal on the receiving side, a block is formed and issued to the channel according to the same rule.

Структура кодовых слов представлена на фиг.1. При этом первые k₂ символов получают от источника, следующие k₁-k₂ повторяют символы от предыдущего вектора

на позициях от k₁+1 до k₀. Эти символы задают номер смежного класса кода А1, которому принадлежит ранее переданный вектор. Символы с номерами k₁+1 до k₀

заполняются нулями, для того чтобы при кодировании вектор у принадлежал коду A₁. Сформированное кодовое слово выдается в канал.If errors are detected on the receiving side during the CRC check, the received block is stored and the NACK error detection signal is sent to the transmitting side via the OS channel. On this signal, the encoder on the receiving side generates the next vector.

The structure of the code words is shown in Fig.1. In this case, the first k ₂ characters are received from the source, the next k ₁ -k ₂ repeat the characters from the previous vector

at positions from k ₁ +1 to k ₀ . These symbols specify the number of the adjacent A1 code class to which the previously transmitted vector belongs. Symbols with numbers k ₁ +1 to k ₀

are filled with zeros so that when encoding the vector y belongs to the code A ₁ . The generated code word is issued to the channel.

в код A₁. Затем производится декодирование с исправлением ошибок с последующей проверкой. Если ошибка не обнаружена, получателю выдаются k₁ информационных символов из накопителя

, а затем k2 символов из накопителя

On the receiving side, error detection is performed in the A ₀ code. If it is decided that there are no errors, the information symbols are decoded, the coset number is used to determine the generatrix of the coset for translation.

to code A ₁ . Then decoding with error correction is performed, followed by verification. If no error is found, the recipient receives k ₁ information characters from the accumulator

and then k2 characters from the drive

повторно. При идущих подряд запросах повторение

продолжается поочередно.If an error occurs after receiving block y, the receiver, on request, sends

again. For consecutive requests, repetition

continues in sequence.

Thus, due to the use of error-correcting code A ₁ during re-interrogation, the number of re-interrogations is reduced, while the structure of nested codes allows you to additionally transmit new characters in code words of length n during the request and limit the number of stored code vectors.

In practical implementation, it is advisable to use A ₀ as a code without redundant code with parameters k ₀ =n and d=1. In this case, error detection can be carried out using a cyclic code - (CRC). This creates an opportunity to apply the method in existing communication systems without interfering with the hardware, due to the implementation at the level of formation of the payload of the packet. Simultaneously performing CRC-16 or CRC-32 checks provides a low probability of undetected errors.

The implementation of the method is explained by the following description of the algorithms for the operation of the receiving and transmitting parts of the system.

on the transmitting side.

и счетчиком поступающих из обратного канала сигналов NACK -N в начале передачи

Передатчик ожидает поступления сигнала ОС (фиг.1). Будем полагать, что формирование первого кодового блока происходит по поступлению АСК от приемника. Если ОС=АСК, то

и от источника поступает набор информационных символов

u_n. Часть символов сохраняют в накопителе

производят кодирование

Результат кодирования сохраняют в накопителе

Производят кодирование циклическим помехоустойчивым кодом, образуя набор CRC

Передают в канал для приемника последовательность символов из накопителя

с добавленным CRC:

После передачи пакета приемник переходит в ожидание сигнала ОС. Если принят сигнал АСК, то последовательность действий, описанная в предыдущем абзаце, повторяется.To implement the method, the transmitting side must have two drives for storing code vectors of length

and a counter of NACK signals coming from the reverse channel -N at the beginning of transmission

The transmitter is waiting for the OS signal (figure 1). We will assume that the formation of the first code block occurs upon receipt of an ACK from the receiver. If OS=ACK, then

and a set of information symbols comes from the source

u _n . Part of the characters are stored in the drive

produce coding

The encoding result is stored in the drive

Encoding is performed with a cyclic error-correcting code, forming a CRC set

A sequence of symbols from the accumulator is transmitted to the channel for the receiver

with added CRC:

After the packet is transmitted, the receiver switches to waiting for the OS signal. If an ACK signal is received, then the sequence of actions described in the previous paragraph is repeated.

При этом от источника получают k₂ символов

Этот набор вместе с сохраненным

и набором нулевых символов длины n-k₁ используют для кодирования:If a NACK signal is received, then the counter of the number of retries is increased by one: N=N+1 and the value of N is checked. If N=1 (the first NACK after the previous ACK), then a new code vector is formed

In this case, k ₂ symbols are received from the source

This set, along with the saved

and a set of zero characters of length nk ₁ is used for encoding:

. Вектор

теперь принадлежит коду А₁. Для него вычисляют

и передают

в канал на приемную сторону, переходят к ожиданию очередного сигнала ОС. Структура

изображена на фиг.2.

. Vector

now belongs to code A ₁ . For him calculate

and pass

into the channel on the receiving side, proceed to waiting for the next OS signal. Structure

shown in Fig.2.

для четного и нечетного значений N соответственно, до момента получения ACK или исчерпания лимита повторений (на фиг.1 не показано).If the number of serial NACK signals received over the OS channel is N≠1, then blocks are transmitted in turn

for even and odd values of N, respectively, until an ACK is received or the repetition limit is exhausted (not shown in FIG. 1).

On the receiving side

В зависимости от количества последовательных переспросов N содержимое

помещается в накопители

и

для обработки переданных по каналу блоков

. В начальный момент N=0, накопители пусты, ожидают прием пакета. Последовательность действий на приемной стороне поясняется на фиг.3.The packet received from the channel enters the drive

Depending on the number of consecutive retries N, the content

placed in storage

and

for processing blocks transmitted over the channel

. At the initial moment N=0, the drives are empty, waiting for the packet to be received. The sequence of actions on the receiving side is illustrated in Fig.3.

восстанавливают сообщение из n символов

, используя преобразование, обратное кодированию. По обратному каналу при этом передают сигнал подтверждения ОС=ACK, N=0, накопителя очищаются, переходят к приему очередного сообщения из канала.If no errors were found in the packet, CRC(Z)=Ok, by

recover a message of n characters

, using the inverse encoding transformation. At the same time, the confirmation signal OS=ACK, N=0 is transmitted over the reverse channel, the drives are cleared, and they proceed to receive the next message from the channel.

формируют сигнал OC=NACK, счетчик количества запросов меняет состояние N=N+1, переходят к приему очередного сообщения из канала.If errors are found in the packet, the received block is stored in the drive

form the signal OC=NACK, the counter of the number of requests changes state N=N+1, proceed to receive the next message from the channel.

приемника уже содержится ранее принятый пакет с обнаруженным сочетанием ошибок. Принятый пакет сохраняют в накопителе

подвергают проверке на CRC и, если ошибок не обнаружено, восстанавливают символы, определяющие номер смежного класса

и набор информационных символов

используя обратимость операции кодирования. Если проверка CRC пакета

привела к обнаружению ошибки, то производят декодирование

в коде A₁ с исправлением ошибок и последующей проверке CRC. При удачной попытке исправления ошибок восстанавливают

Если попытка декодирование не привела к исправлению ошибок, то в канал ОС отправляют NACK, увеличивают значение счетчика N=N+1 и переходят к ожиданию очередного пакета из канала.For any odd N (N=1mod2), in the drive

receiver already contains a previously received packet with a detected pattern of errors. The received packet is stored in the drive

subjected to a CRC check and, if no errors are found, restore the characters that determine the number of the adjacent class

and a set of information symbols

using the reversibility of the encoding operation. If the packet CRC check

led to the detection of an error, then decoding is performed

in code A ₁ with error correction and subsequent CRC check. If the error correction attempt is successful, restore

If the decoding attempt did not lead to error correction, then NACK is sent to the OS channel, the counter value N=N+1 is increased, and the next packet is waiting from the channel.

успешно декодирован (или принят без ошибок), то полученные в нем символы

задают номер смежного класса и позволяют выполнить декодирование кодового вектора из накопителя

в коде A₁ с исправлением ошибок и проверкой CRC.If

successfully decoded (or received without errors), then the characters received in it

specify the number of the adjacent class and allow decoding of the code vector from the accumulator

in code A ₁ with error correction and CRC check.

в коде A₁ успешно, то выделяют k₁ символов, объединяют с n-k₁ символами

(напомним, что n-k1=k1-k2) и k₂ символами, полученными из вектора

Получателю выдают n+k₂ символов, передают в обратный канал ОС=ACK, счетчик обнуляют N=0, накопители очищают Х=0 Y=0, переходят к ожиданию следующего пакета. Если же декодирование

в коде A₁ не успешно, то в канал ОС отправляют NACK, увеличивают значение счетчика N=N+1 и переходят к ожиданию очередного пакета из канала.If decoding

in code A ₁ is successful, then k ₁ symbols are extracted, combined with nk ₁ symbols

(recall that n-k1=k1-k2) and k ₂ characters obtained from the vector

The recipient is given n+k ₂ symbols, transmitted to the reverse channel OS=ACK, the counter is reset to N=0, the drives clear X=0 Y=0, proceed to wait for the next packet. If the decoding

in code A ₁ is not successful, then NACK is sent to the OS channel, the counter value N=N+1 is increased and the next packet is waiting for the next packet from the channel.

помещают в накопитель

производят проверку CRC. Если ошибок не обнаружено, то выделяют номер смежного класса

и декодируют

коде А₂ с последующей проверкой CRC. Если декодирование в коде А₂ успешно, получателю выдают n+k₂ символов, передают в обратный канал ОС:=ACK, устанавливают N=0, Х=0, Y=0 и переходят к ожиданию следующего пакета.For any even N≠0 (N=0 mod2) the received packet

placed in storage

perform a CRC check. If no errors are found, then the number of the adjacent class is allocated

and decode

code A ₂ followed by a CRC check. If the decoding in code A ₂ is successful, the receiver is given n+k ₂ symbols, transmitted in the reverse channel OS:=ACK, set N=0, X=0, Y=0 and proceed to wait for the next packet.

вычисление CRC привело к обнаружению ошибки, то в канал ОС отправляют NACK, увеличивают значение счетчика N=N+1 и переходят к ожиданию очередного пакета из канала.If in

calculation of the CRC led to the detection of an error, then a NACK is sent to the OS channel, the counter value N=N+1 is increased, and the next packet is waiting for the channel.

Claims (1)

A method for transmitting messages in a system with decisive feedback, which consists in the fact that on the transmitting side the sequence of information symbols is divided into blocks of a certain length, which are encoded with an error-correcting code and transmitted to the communication channel, on the receiving side of the communication system they are decoded with error detection, when errors are detected the received code block is stored, a signal about the detected error is generated, and this signal is transmitted via the feedback channel to the transmitting side, where additional (redundant) symbols are formed and transmitted in the code block, then after receiving these symbols, the previously stored vector is re-decoded, characterized in that that an error-correcting code is used for transmission, which forms a system of nested codes, all transmitted code blocks are of equal length, the code block transmitted on request includes as additional information the number of the adjacent code class to which the stored code vector belongs, as well as additional significant information symbols, decode the transmitted vector and the stored vector with correction and error detection in the code with greater redundancy, when errors are detected, code vectors are transmitted in turn and re-decoding with error correction is performed.

Images