CN115225202A

CN115225202A - Cascade coding and decoding method

Info

Publication number: CN115225202A
Application number: CN202210192714.1A
Authority: CN
Inventors: 王中风; 杨蕾; 崔航轩
Original assignee: Nanjing University
Current assignee: Nanjing University
Priority date: 2022-03-01
Filing date: 2022-03-01
Publication date: 2022-10-21
Anticipated expiration: 2042-03-01
Also published as: CN115225202B

Abstract

The method for cascade coding and decoding adopts RS codes as external codes and adopts a multilayer coding idea as internal codes by utilizing the characteristic that low bits of each transmission symbol are easier to make mistakes than high bits after transmission data are modulated. The outer code of the cascade coding method adopts RS code, interweaves RS code word data, divides the interweaved data into bit data with preset quantity groups, codes each group of bit data by adopting code words with different code rates to obtain multi-layer sub-code word data with preset quantity groups, and modulates to obtain modulated data; wherein, the length of each group of multi-layer sub-code word data is consistent and respectively corresponds to different bits in the modulation data. The cascade coding method uses shorter code length to realize higher coding gain under the condition of certain code rate; the cascade decoding uses the low-order decoding result to assist in demodulating and decoding the high-order data in the channel receiving data, obtains higher decoding performance, and meets the decoding performance requirement of the Ethernet with the speed of over 400 Gb/s.

Description

Cascade coding and decoding method

Technical Field

The present application relates to the field of digital communication coding and decoding technologies, and in particular, to a method for cascade coding and decoding.

Background

In a digital communication system, a source generates data and transmits the data through a channel. However, the channel may cause interference to the transmitted data, so that the sink cannot correctly receive the transmitted data. Generally, a Forward Error Correction (FEC) method is used to correct errors generated during transmission, that is, a data transmitting end encodes data generated by an information source, a data receiving end detects and corrects the errors by using a corresponding decoding algorithm, and the decoded data is transmitted to an information sink.

According to the ethernet latest standard, two RS codes (Reed-Solomon, codes) are applied as FEC schemes to the ethernet sublayer, an RS (528, 514) code corresponding to the KR4 FEC scheme in the ethernet standard and an RS (544, 514) code corresponding to the KP4 FEC scheme in the ethernet standard, respectively.

However, as the transmission rate of ethernet is higher and higher, the ethernet standard above 400Gb/s is being studied, and the decoding performance requirement for the FEC scheme is also higher and higher. On the premise of a certain code rate, to increase the coding gain of the FEC scheme, the code length of the codeword needs to be increased, but the increase of the code length increases the complexity of the communication system and occupies too many hardware resources. Therefore, the single RS codeword FEC scheme described above cannot meet the decoding performance requirements for ethernet applications beyond 400 Gb/s.

Disclosure of Invention

In order to solve the problem that the single-codeword FEC method cannot meet the decoding performance requirements of ethernet applications beyond 400Gb/s, the present application provides a concatenated coding and decoding method through the following aspects.

A first aspect of the present application provides a concatenated coding method, including:

receiving information source data transmitted by an information source;

RS coding is carried out on the information source data to obtain RS code word data;

interweaving the RS code word data according to a preset interweaving depth to obtain interweaved data;

grouping the interleaved data to obtain bit data of a preset number of groups; the preset number is the number of bits corresponding to one symbol in the preset modulation mode, and each group of bit data corresponds to different bits in the preset modulation mode;

carrying out multilayer coding on the bit data of a preset number of groups to obtain multilayer code word data; the multi-layer code word data comprises a preset number of groups of multi-layer sub-code word data, and the length of each group of multi-layer sub-code word data is consistent; the check bit length in the multi-layer sub-code word data corresponding to the most significant bit in the preset modulation mode is smaller than the check bit length in the multi-layer sub-code word data corresponding to the least significant bit in the preset modulation mode;

and modulating the multilayer code data according to a preset modulation mode to obtain modulation data for channel transmission.

Optionally, when the preset modulation mode is PAM4, the preset number of groups of bit data includes a group of high-order bit data and a group of low-order bit data;

the multi-layer coding of the bit data of the preset number of groups to obtain multi-layer codeword data includes:

coding the low-order bit data according to a first code rate to obtain low-order multilayer subcode digital data;

coding the high-order bit data according to a second code rate to obtain high-order multilayer subcode word data;

the first code rate is smaller than the second code rate, the high-order multi-layer sub-codeword data corresponds to the most significant bit in the PAM4 modulation mode, and the low-order multi-layer sub-codeword data corresponds to the least significant bit in the PAM4 modulation mode.

the multi-layer coding is performed on the bit data of the preset number of groups to obtain multi-layer codeword data, and the multi-layer codeword data includes:

coding the low-order bit data according to a third code rate to obtain low-order multilayer subcode word data;

directly outputting the high-order bit data without encoding to obtain high-order multilayer subcode word data;

the high-order multi-layer sub-codeword data corresponds to the most significant bit in the PAM4 modulation mode, and the low-order multi-layer sub-codeword data corresponds to the least significant bit in the PAM4 modulation mode.

A second aspect of the present application provides a concatenated decoding method for decoding modulation data obtained by the concatenated coding method according to the first aspect of the present application; the cascade coding method comprises the following steps:

executing a first demodulation process on channel receiving data according to a preset modulation mode to obtain a first quantity of groups of low-order demodulation data; decoding the low-order demodulated data to obtain low-order bit data

Processing the channel receiving data with the assistance of low-order bit data to obtain a second number of groups of high-order bit data; the first number plus the second number is equal to a preset number, and the preset number is the number of bits corresponding to one symbol in a preset modulation mode;

performing de-interleaving processing on the high-order bit data and the low-order bit data to obtain de-interleaved data;

RS decoding is carried out on the deinterleaved data to obtain RS decoded data;

the RS decoded data is transmitted to the sink.

Optionally, when the preset modulation mode is PAM4, both the first number and the second number are equal to 1; the lower bit data corresponds to the least significant bit of the modulation data, and the upper bit data corresponds to the most significant bit of the modulation data;

the processing the channel receiving data with the assistance of the low-order bit data to obtain the high-order bit data includes:

when the most significant bit in the modulation data is not coded in the multi-layer coding process, using the low-order bit data to assist in executing a second demodulation process on the most significant bit of the channel receiving data in a hard decision mode to obtain high-order bit data;

when the most significant bit in the modulation data is coded in the multi-layer coding process, the low-bit data is used for assisting in executing a third demodulation process on the most significant bit of the channel receiving data in a hard decision mode or a soft decision mode, and the demodulation result is decoded to obtain high-bit data; when the third demodulation process is executed by adopting hard decision, decoding the demodulation result by adopting a hard decoding mode; and when the third demodulation process is executed in a soft decision mode, decoding the demodulation result in a soft decoding mode.

Optionally, before sending the RS decoded data to the sink, the cascade decoding method includes:

performing iterative decoding on the RS decoded data to obtain RS iterative decoded data;

the RS iterative decoding data is sent to an information sink;

the iterative decoding of the RS decoded data to obtain RS iterative decoded data includes:

interweaving RS decoding data to obtain iterative input data; wherein the iterative input data comprises low-order iterative input data and high-order iterative input data;

decoding the low-order iteration input data to obtain low-order iteration bit data;

when the most significant bit in the modulation data is not coded in the multi-layer coding process, using low-order iteration bit data to assist in executing a second demodulation process on high-order iteration input data in a hard decision mode to obtain high-order iteration bit data;

when the most significant bit in the modulation data is coded in the multi-layer coding process, the low-order iterative bit data is used for assisting to execute a third demodulation process on the high-order iterative input data in a hard decision mode or a soft decision mode, and the demodulation result is decoded to obtain high-order iterative bit data; when the third demodulation process is executed in a hard decision mode, decoding the demodulation result in a hard decoding mode; when the third demodulation process is executed by adopting a soft decision mode, decoding the demodulation result by adopting a soft decoding mode;

performing de-interleaving processing on the high-order iteration bit data and the low-order iteration bit data to obtain de-interleaved iteration data;

RS decoding is carried out on the deinterleaved iterative data to obtain RS iterative decoding data;

and transmitting the RS iterative decoding data to the information sink.

Optionally, when the low-order iterative input data is decoded in a hard decoding manner, the check bits in the low-order iterative input data are set as the check bits in the low-order bit data.

Optionally, when the low-order iterative input data is decoded in a soft decoding manner, the maximum likelihood ratio of the information bits in the low-order iterative data is set to be most reliable, and the maximum likelihood ratio of the check bits in the low-order iterative input data is set to be the maximum likelihood ratio in the low-order demodulated data.

Optionally, a hard decision method is adopted when the first demodulation process is executed on the channel received data according to the PAM4 modulation method, so as to obtain low-order demodulated data;

and when the low-order demodulation data are decoded, a hard decoding mode is adopted to obtain the low-order bit data.

Optionally, a soft decision manner is adopted when a first demodulation process is executed on the channel received data according to a PAM4 modulation manner, so as to obtain low-order demodulated data;

and when decoding the low-order demodulation data, a soft decoding mode is adopted to obtain the low-order bit data.

The application provides a cascade coding and decoding method. The method for cascade coding and the corresponding cascade decoding method have the advantages that the characteristic that the low order bit of each transmission symbol is easier to make mistakes than the high order bit after the transmission data is modulated is utilized, and the method for cascade coding and the corresponding method for cascade decoding are provided, wherein the RS code is adopted by an outer code, and the multilayer coding idea is applied to an inner code. The outer code of the cascade coding method adopts RS code, interweaves RS code word data, then divides the interweaved data into bit data with preset number groups, codes each group of bit data by adopting code words with different code rates to obtain multilayer sub-code word data with preset number groups, and then modulates to obtain modulated data; wherein, the length of each group of multi-layer sub-code word data is consistent and respectively corresponds to different bits in the modulation data. The cascade coding method realizes higher coding gain by using shorter code length under the condition of certain code rate.

When decoding, firstly demodulating and decoding the low order bits in the channel receiving data to obtain low order bit data; using low-order bit data to assist in processing high-order bits in channel receiving data to obtain high-order bit data; and de-interleaving and RS decoding the high-order bit data and the low-order bit data to obtain RS decoded data and sending the RS decoded data to an information sink. The FEC scheme using the cascade coding and decoding method provided by the application can meet the decoding performance requirement of the Ethernet application exceeding 400 Gb/s.

Drawings

FIG. 1 is a block diagram of a communication system using two codes to form a concatenated code;

fig. 2 is a schematic flowchart of a concatenated coding method according to an embodiment of the present application;

fig. 3 is a schematic diagram of a dividing manner of grouping interleaved data in a concatenated coding method according to an embodiment of the present application;

fig. 4 is a schematic diagram illustrating an example of MLC encoding in a concatenated encoding method according to an embodiment of the present application;

fig. 5 is a constellation diagram example of modulated data in the 16QMA modulation scheme;

fig. 6 is a schematic flowchart of a concatenated decoding method according to an embodiment of the present application;

fig. 7 is a schematic diagram illustrating an MLC encoding/decoding process in a concatenated decoding method according to an embodiment of the present application;

fig. 8 is a schematic diagram illustrating a process of using low-order bit data to assist in performing hard-decision demodulation on channel received data in a concatenated decoding method according to an embodiment of the present application;

fig. 9 is a schematic diagram illustrating a process of processing channel received data with assistance of low-order bit data in a concatenated decoding method to obtain high-order bit data according to an embodiment of the present application;

FIG. 10 is a schematic diagram showing comparison of simulation results of decoding performance of the experimental group 1, the comparative group 1 and the comparative group 2;

FIG. 11 is a schematic diagram showing comparison between the decoding performance simulation results of experiment set 2 and experiment set 3;

fig. 12 is a schematic diagram showing comparison of simulation results of decoding performance of the experimental group 1, the experimental group 4, the experimental group 5, and the comparative group 1.

Detailed Description

For the purpose of explaining the technical scheme of the application, the concept of concatenated coding to which the present application relates will be briefly introduced.

The concatenated code is composed of two or more simple codes, and fig. 1 exemplarily shows a structural diagram of a communication system in which two codes constitute the concatenated code. As shown in fig. 1, the concatenated code includes an outer code (typically non-binary) and an inner code (typically binary), which are two separate codes that are transmitted serially over a channel. The method comprises the steps of firstly carrying out outer code coding on data sent by an information source, then carrying out inner code coding after interleaving, then carrying out channel transmission, carrying out two-stage decoding on a decoding end by a cascade code, firstly carrying out inner code decoding, and then carrying out outer code decoding after de-interleaving.

Referring to fig. 2, a flowchart of a concatenated coding method according to a first embodiment of the present application is shown. The concatenated coding method includes steps 11 to 16.

And 11, receiving source data transmitted by a source. The source is a generation end and a transmission end of data in the communication system, and converts various messages into original electric signals, which are referred to as source data in this application.

And step 12, performing RS coding on the source data to obtain RS code word data.

In this embodiment, the outer code employs an RS codeword. In the latest ethernet standard, two RS codewords are used to encode the source data. The RS code is transmitted in the form of symbol, one is RS (528, 514) code, corresponding to KR4 FEC scheme in ethernet standard; wherein, one RS code word has 528 code elements and 514 information code elements. The other is RS (544, 514) code, wherein, one RS code word has 544 symbols and 514 information symbols; corresponding to the KP4 FEC scheme in the ethernet standard. 1 symbol is 1 symbol (symbol number), and 1 symbol is =10bits.

And step 13, interleaving the RS code word data according to a preset interleaving depth to obtain interleaved data.

And carrying out symbol-level row-column interleaving on the RS code word data, wherein the preset interleaving depth is I, and I can be any positive integer. However, the delay increases with the increase of I, and in practical application, an appropriate I is configured according to the requirements of delay, throughput rate and the like.

And 14, grouping the interleaved data to obtain bit data of a preset number of groups. The preset number is the number of bits corresponding to one symbol in a preset modulation mode, and each group of bit data corresponds to different bits in the preset modulation mode.

Step 15, carrying out multilayer coding on the bit data of a preset number of groups to obtain multilayer code word data; the multi-layer code word data comprises a preset number of groups of multi-layer sub-code word data, and the length of each group of multi-layer sub-code word data is consistent; the length of the check bits in the multi-layer sub-codeword data corresponding to the most significant bits in the preset modulation mode is smaller than the length of the check bits in the multi-layer sub-codeword data corresponding to the least significant bits in the preset modulation mode.

In this embodiment, the execution of steps 14-15 is described by using a PAM4 (4 Pulse Amplitude Modulation, fourth generation Pulse Amplitude Modulation) Modulation scheme and an RS (544, 514) code as an outer code. In PAM4 modulation, two bits are modulated into one symbol, with a preset number equal to 2. Dividing the interleaved data into N parts, each part comprises 2k bit data, and dividing the 2k bit data into high-order bit data (k in total) ₁ One) and low-order bit data (total k) ₂ One); wherein 2k = k ₁ +k ₂ ，k ₁ ＞k ₂ (ii) a That is, the interleaved data is divided into 2 groups, the first group containing N parts of k ₁ Data of one Bit corresponding to data on Most Significant Bit (MSB) in PAM4 modulation, and a second group containing N parts of k ₂ One Bit data, corresponding to data on the Least Significant Bit (LSB) in PAM4 modulation.

In practical applications, there are various ways to divide 2k bit data into high bit data (total k) ₁ One) and low-order bit data (total k) ₂ One). FIG. 3 shows the present embodiment for dividing 2k bit data into high-order bit data (k total) ₁ One) and low-order bit data (total k) ₂ One example of a division manner of). As shown in fig. 3, for 2k bitsFirst 2k in data ₁ Dividing the bit data according to the odd-even sequence, wherein the odd number is high-order bit data, and the even number is low-order bit data; rear (k) ₂ -k ₁ ) The individual data is high order bit data. In other examples, the partitioning may be performed in other orders as well, as long as 2k = k is satisfied ₁ +k ₂ And k is ₁ ＞k ₂ That is, the partitioning method is not particularly limited in the present application.

The high-order bit data and the low-order bit data are subjected to Multi-level coding (MLC) to obtain Multi-level codeword data (referred to as MLC codeword data in this application). In this embodiment, taking PAM4 as an example of the preset modulation scheme, the multi-layer codeword data includes two sets of MLC sub-codeword data. Selecting a BCH code as a composition code of the MLC code, namely encoding high-order bit data and low-order bit data by adopting BCH codes with different code rates to obtain two groups of MLC subcode codeword data with the same length; the length of one MLC sub-codeword data is n, and the length of one MLC codeword data is 2n.

Referring to fig. 4, an exemplary diagram of MLC encoding in this embodiment is shown. As can be seen from the calculation formula of the code rate (code rate = information bit length/code length), when the code rate and the code length are fixed, the bit data length of the information bit is also fixed. In FIG. 4, k is k ₁ And k ₂ Average value of (a). In practical application, according to the design requirements of an actual communication system, the code rate and the code length 2n of the MLC code word are combined to deduce the information bit length 2k of the MLC code word, and then suitable k is further deduced ₁ And k ₂ 。

In one implementation manner, in the MLC coding, a BCH code with a first code rate is used to code low-order bit data, and a BCH code with a second code rate is used to code high-order bit data, so as to obtain low-order multi-layer sub-codeword data and high-order multi-layer sub-codeword data, respectively, where the first code rate is smaller than the second code rate, that is, the error correction capability of the low-order multi-layer sub-codeword data is higher than the error correction capability of the high-order multi-layer sub-codeword data. Illustratively, the low-order bit data is encoded by using a BCH (360, 320) code, and the error correction capability is 4; the high-order bit data is encoded by using BCH (360, 340) code, and the error correction can be carried outThe force is 2. Correspondingly, n =360,k ₁ ＝320，k ₂ ＝340。

In another implementation manner, in the MLC coding, a BCH code with a third code rate is used to code low-order bit data, so as to obtain low-order multi-layer subcode word data; the high-order bit data is not coded, and is directly output to obtain high-order multilayer subcode data, namely the high-order multilayer subcode data has no error correction capability. Illustratively, the low order bit data is encoded using a BCH (360, 320) code, corresponding to k ₁ ＝320，k ₂ ＝360，n＝360。

In this embodiment, since the code length of the outer code is different from that of the inner code, the interleaved data is divided into N parts, and each part is encoded. And N is determined according to the data length of the RS code words, the preset interleaving depth and the length of the information bits of the coding code words selected in the multi-layer coding in the step 15. In an exemplary manner, the first and second electrodes are,

wherein, the preset interleaving depth is set as 2, the outer code adopts RS (544, 514) code, the data length of MLC code word of the inner code is equal to 360 (corresponding to the high bit data without error correction capability) +320 (corresponding to the low bit data with error correction capability of 4, BCH (360, 320) code is adopted).

It should be noted that, in the above example, the MLC codes use BCH codes as the constituent codes, and other code words may also be used as the constituent codes, for example, LDPC codes or Polar codes are used as the constituent codes of MLC codes.

And step 16, modulating the multilayer code word data according to a preset modulation mode to obtain modulation data for channel transmission. In this embodiment, taking the PAM4 modulation scheme as an example, corresponding bits of the higher-order multi-layer sub-codeword data (corresponding to the MSB in the modulated data) and the lower-order multi-layer sub-codeword data (corresponding to the LSB in the modulated data) are modulated into one sign bit in the modulated data by PAM 4. Table 1 shows the correspondence between the multi-layer codeword data and the modulation data in the PAM4 modulation scheme.

Table 1: modulating the multi-layer code word data according to PAM4

(MSB，LSB)	(0,0)	(0,1)	(1,0)	(1,1)
					After PAM4 modulation	-3	-1	1	3

As can be seen from the data format of PAM4 modulated data in table 1, when the modulated data is less than 0, the corresponding highest bits are all 0, and when the modulated data is greater than 0, the highest bits are all 1. It can be seen that the highest bit has the greatest effect on the value of the modulated data. Therefore, in the encoding process, the encoding modes with different code rates are adopted for the bit data corresponding to the least significant bit and the most significant bit. In the decoding process, even if the least significant bit is decoded incorrectly, the value corresponding to the most significant bit can be judged only according to whether the modulation data is greater than 0 or less than 0.

Exemplarily, referring to fig. 5, it is a schematic diagram of an example of a constellation diagram of 16QAM modulated data. Taking the un-bolded two-bit decoding as 00 as an example, the low-bit auxiliary high-bit demodulation process is shown in table 2.

Table 2:16QAM Low-order auxiliary high-order demodulation process (taking the low order as 00 for example)

Not thickened with two bits 00	Re<1,Im>-j	Re<1,Im<-j	Re>1,Im>-j	Re>1,Im<-j
					Two bit demodulation result is thickened	10	11	00	01

It should be noted that, when data is modulated, gray codes are generally used to reduce the possibility of errors and improve the decoding performance. However, in the encoding method provided in the present application, if gray codes are used for modulation during modulation, when the lower-order bit data is used to assist demodulation and decoding of the higher-order bit data, the higher-order bit data has a high possibility of error even if the lower-order bit data is decoded correctly. Therefore, in order to make the decoding result of the high-order bit data as reliable as possible, the gray code is not used in the modulation process when the cascade coding method is used for designing a communication system.

The embodiment provides a cascade coding method, which comprises the steps of receiving source data sent by a source; RS coding is carried out on the information source data to obtain RS code word data; interweaving the RS code word data according to a preset interweaving depth to obtain interweaved data; grouping the interleaved data to obtain bit data of a preset number of groups; the preset number is the number of bits corresponding to one symbol in the preset modulation mode, and each group of bit data corresponds to different bits in the preset modulation mode; carrying out multilayer coding on the bit data of a preset number of groups to obtain multilayer code word data; the multi-layer code word data comprises a preset number of groups of multi-layer sub-code word data, and the length of each group of multi-layer sub-code word data is consistent; the check bit length in the multi-layer sub-code word data corresponding to the most significant bit in the preset modulation mode is smaller than the check bit length in the multi-layer sub-code word data corresponding to the least significant bit in the preset modulation mode; and modulating the multilayer code word data according to a preset modulation mode to obtain modulation data for channel transmission.

The cascade coding method utilizes the characteristic that the low bit of each transmission symbol is easier to make an error than the high bit of the transmission data after modulation, internal codes with different error correction capabilities are respectively adopted for the low bit and the high bit of the modulation symbol, different error correction capabilities are provided for different bits of the modulated symbol, and under the condition of certain code rate, higher coding gain is realized by using shorter code length.

In the first embodiment, the preset modulation scheme is PAM 4. The preset modulation mode in the concatenated coding method provided in the above embodiment may also be another modulation mode.

Illustratively, when the preset modulation mode is 16QAM, one symbol in the 16QAM modulation data corresponds to four bits, and the interleaved data is grouped to obtain four groups of bit data, where one group of bit data corresponds to one bit in the 16QAM modulation data. The constituent codes in the MLC codeword may be selected to be BCH (360, 350), BCH (360, 330), or BCH (360, 350), BCH (360, 340), BCH (360, 330), respectively, corresponding to the order of the modulated data from the upper bits to the lower bits. As long as the following two conditions are satisfied: after MLC coding, the lengths of the MLC sub-codeword data are consistent, and the error correction capability of the MLC sub-codeword data corresponding to the least significant bit is higher than that of the MLC sub-codeword data corresponding to the most significant bit.

Corresponding to the foregoing embodiments of the concatenated coding method, the present application further provides an embodiment of a concatenated decoding method. Referring to fig. 6, a schematic flowchart of a concatenated decoding method according to a second embodiment of the present application is shown. The concatenated decoding method comprises steps 21-26.

Step 21, a first demodulation process is performed on the channel received data according to a preset modulation mode, so as to obtain a first quantity of groups of low-order demodulated data. The channel received data refers to data obtained by transmitting modulated data obtained by the concatenated coding method provided in the first embodiment of the present application through a channel and adding a channel noise signal.

In this embodiment, the first demodulation process is performed in a preset modulation scheme corresponding to the concatenated coding method. The first demodulation process is to demodulate low-order data in the channel receiving data to obtain low-order demodulation data. When the preset modulation mode in the cascade coding method uses PAM4 modulation, demodulation is performed according to the PAM4 modulation mode in a corresponding first demodulation process, the first number is equal to 1, and low-order demodulation data corresponds to the least significant bit in PAM4 modulation data; when the preset modulation mode in the cascade coding method uses 16QAM modulation, the corresponding first demodulation process also needs to demodulate according to the 16QAM modulation mode, the first number is equal to 2, and the low-order demodulation data corresponds to the least significant bit and the second least significant bit in the 16QAM modulation data.

In step 21, the lower demodulated data corresponds to the lower part of the multi-layer sub-codeword data in the concatenated coding method. Illustratively, when the preset modulation mode is 16QAM, the low-order demodulation data corresponds to two sets of low-order multi-layer sub-codeword data, and when the preset modulation mode is PAM4, the low-order demodulation data corresponds to one set of multi-layer sub-codeword data corresponding to the least significant bit.

And step 22, decoding the low-order demodulated data to obtain low-order bit data.

In one implementation, a first demodulation process may be performed on the PAM4 modulated data in a hard decision manner to obtain lower-order demodulated data. Correspondingly, the low-order demodulated data is decoded in a hard decoding mode to obtain low-order bit data.

In another implementation manner, a first demodulation process may be performed on PAM4 modulated data in a soft decision manner to obtain low-order bit data; correspondingly, the low-order demodulated data is decoded according to a soft decoding mode to obtain low-order bit data.

The soft decision and soft decoding mode has good decoding performance and low error rate, but the complexity is slightly high, and a channel is required to provide soft information. The complexity is low by adopting a hard decision and hard decoding mode, the decoding performance is lower than that of a soft decision and soft decoding mode, but a channel is not required to provide soft information, and the power consumption of the equalizer is greatly reduced. In practical applications, different decision manners may be selected to perform the first demodulation process according to requirements of specific application scenarios.

In this embodiment, when the constituent code of the MLC coding in the concatenated coding method selects the BCH code, correspondingly, when the concatenated decoding is performed, the BM (Berlekamp-Massey) algorithm may be used for the hard decoding of the BCH code, and the Chase-II algorithm may be used for the soft decoding.

Step 23, using the low-order bit data to assist in processing the channel receiving data to obtain a second number of groups of high-order bit data; the first number plus the second number is equal to a preset number, and the preset number is the number of bits corresponding to one symbol in the preset modulation mode.

In the present embodiment, the lower-order decoding result of the channel reception data is used to assist in high-order demodulation of the channel reception data. Referring to fig. 7, a schematic diagram of the MLC encoding and decoding process is shown. As shown in fig. 7, taking a PAM4 modulation method as an example, low-order demodulation is performed on channel received data to obtain low-order demodulated data, and then the low-order demodulated data is decoded to obtain low-order bit data; and demodulating the channel receiving data by using the low-order bit data to obtain high-order demodulated data, and decoding the high-order demodulated data to obtain high-order bit data. Wherein the AWGN channel in fig. 7 refers to a white gaussian noise channel, which is used to simulate real channel noise.

Referring to fig. 8, a schematic diagram of a process of demodulating the modulated data with assistance of low-order bit data is shown, where the demodulation method shown in fig. 8 adopts a hard decision manner to demodulate, so as to obtain binary high-order demodulated data; here, noise _ symbol in fig. 8 represents modulated data received by a channel. When the low-order bit data is 0 and the channel receiving data is more than-1, the corresponding high-order demodulation data is 1; when the low-order bit data is 0 and the channel receiving data is less than-1, the corresponding high-order demodulation data is 0; when the low-order bit data is 1 and the channel receiving data is more than 1, the corresponding high-order demodulation data is 1; when the low-order bit data is 1 and the channel receiving data is less than 1, the corresponding high-order demodulation data is 0.

When the low-order bit data is used to assist in demodulating the channel received data in a soft decision manner, the obtained high-order demodulated data is soft information, and the specific assisting process is similar to that in fig. 8 and is not described herein again.

In practical application, when the low-order bit data is used for assisting in demodulating the channel receiving data by hard decision, binary high-order demodulated data is obtained, and correspondingly, a hard decoding mode is adopted when the high-order demodulated data is decoded. When the low-order bit data is used for assisting in demodulating the channel receiving data by adopting soft decision, high-order demodulation data of soft information is obtained, and correspondingly, a soft decoding mode is adopted when the high-order demodulation data is decoded.

Referring to fig. 9, a PAM4 modulation scheme is taken as an example, and a workflow diagram of obtaining high-order bit data by processing channel received data with assistance of low-order bit data is given according to whether the most significant bit in the modulated data is encoded in a multi-layer encoding process.

When the most significant bit in the modulation data is not coded in the multi-layer coding process, the decoding is not needed in the cascade decoding process, and the high-bit demodulation data is directly output. When the low-order bit data is used for assisting in demodulating the channel receiving data, a hard decision mode is adopted, and the obtained binary high-order demodulated data is directly output into high-order bit data. In the present embodiment, the above-described demodulation process is referred to as a second demodulation process. At this time, as shown in FIG. 9, step 23 further includes step 231-1, step 232-1 and step 233.

And 231-1, demodulating the channel received data by using low-order bit data in a hard decision mode to obtain binary high-order demodulated data. The specific demodulation process is shown in fig. 8.

Step 232-1, the binary high-order demodulated data is directly output.

In step 233, high order bit data is obtained.

When the most significant bit in the modulated data is encoded in the multi-layer encoding process, the high-order demodulated data needs to be decoded in the cascade decoding process. In this embodiment, the demodulation process for obtaining the high-order demodulated data to be decoded is referred to as a third demodulation process.

And when the third demodulation process is executed in a hard decision mode, decoding the obtained high-order demodulation data in a hard decoding mode. At this time, as shown in fig. 9, step 23 further includes step 231-2, step 232-2 and step 233.

And 231-2, using low-order bit data to assist in demodulating the channel receiving data, and obtaining binary high-order demodulation data in a hard decision mode. The demodulation process is shown in fig. 8.

And step 232-2, decoding the binary high-order demodulated data by adopting a hard decoding mode.

In step 233, high order bit data is obtained.

When the third demodulation process is executed in a soft decision mode, the obtained high-order demodulation data is decoded in a soft decoding mode. At this time, as shown in FIG. 9, step 23 further includes step 231-3, step 232-3 and step 233.

And 231-3, using low-order bit data to assist in demodulating the channel receiving data, and adopting a soft decision mode to obtain high-order demodulation data of soft information.

And step 232-3, decoding the high-order demodulation data of the soft information by adopting a soft decoding mode.

In step 233, high-order bit data is obtained.

In the present embodiment, the execution process of steps 21 to 23 is referred to as MLC decoding process. When the first demodulation process in step 21 adopts a soft decision method, it is called MLC soft decoding. Further, the MLC soft decoding includes using low-order bit data to assist in demodulating the channel received data in both hard decision and soft decision forms. When the first demodulation process in step 22 adopts a hard decision method, it is called MLC hard decoding. Further, the MLC hard decoding involves using low-order bit data to assist in demodulating the channel received data in both hard decision and soft decision forms. That is, the MCL soft decoding includes two forms of low-order soft-decision demodulation + high-order soft-decision demodulation and low-order soft-decision demodulation + high-order hard-decision demodulation; the MLC hard decoding comprises two forms of low-order hard decision demodulation, high-order hard decision demodulation and low-order hard decision demodulation and high-order soft decision demodulation.

And 24, performing deinterleaving processing on the high-order bit data and the low-order bit data to obtain deinterleaved data.

And 25, performing RS decoding on the de-interleaved data to obtain RS decoded data.

And step 26, sending the RS decoding data to a signal sink.

And de-interleaving the MLC decoding result, reducing the MLC decoding result into the sequence of RS code words, performing RS decoding, obtaining RS decoding data and sending the RS decoding data to an information sink. In this embodiment, the RS decoding adopts a hard decoding method.

In this embodiment, the performance of the concatenated decoding method may also be increased by means of iterative decoding, that is, after RS hard decoding is completed, RS decoding data is used as iterative input data, MLC decoding and RS decoding processes are performed again, and an obtained iterative decoding result is sent to an information sink. Specifically comprising steps 31-36.

And step 31, interleaving the RS decoding data to obtain iterative input data. It should be noted that the RS decoded data is not encoded after being interleaved, and the interleaved data is directly used as input data of iterative decoding and is again used as input of the MLC decoding process. The iterative input data includes lower iterative input data and higher iterative input data.

And 32, decoding the low-order iteration input data to obtain low-order iteration bit data. Taking PAM4 modulation as an example, the low-order iterative input data is directly decoded to obtain low-order iterative bit data. Because the RS decoded data does not have information of the parity bits, the discarded parity bits need to be processed in the iterative input data obtained after the RS decoded data is interleaved. In one implementation mode, decoding low-order iterative input data in a hard decoding mode, and setting a check bit in the iterative input data as a check bit of data after first low-order decoding, namely the check bit in the low-order bit data; in another implementation, the low-order iterative input data is decoded by a soft decoding method, the maximum likelihood ratio of the information bits in the low-order iterative data is set as the most reliable, and the maximum likelihood ratio of the check bits in the low-order iterative input data is set as the maximum likelihood ratio in the low-order demodulated data.

And step 33, processing the high-order iteration input data by using the low-order iteration bit data to obtain high-order iteration bit data. Step 33 further includes step 331 and step 332 depending on whether the most significant bits in the modulated data are encoded in the multi-layer encoding process.

And 331, when the most significant bit in the modulation data is not coded in the multi-layer coding process, using the low-order iteration bit data to assist in executing a second demodulation process on the high-order iteration input data in a hard decision mode to obtain high-order iteration bit data.

And 332, when the most significant bit in the modulation data is coded in the multi-layer coding process, using the low-order iteration bit data to assist in executing a third demodulation process on the high-order iteration input data in a hard decision mode or a soft decision mode to obtain high-order demodulation iteration data, and decoding the high-order demodulation iteration data to obtain the high-order iteration bit data. Further, when the third demodulation process is executed in a hard decision mode, decoding the high-order demodulation iteration data in a hard decoding mode; and when the third demodulation process is executed in a soft decision mode, decoding the high-order demodulation iteration data in a hard decoding mode.

In this embodiment, steps 32 to 33 are referred to as an MLC iterative decoding process, where when the low-order iterative input data is decoded by using a hard decoding method, the decoding process is referred to as an MLC hard iterative decoding process; when the low-order iterative input data is decoded by adopting a soft decoding mode, the decoding process is called an MLC soft iterative decoding process. Furthermore, the MLC hard iterative decoding process includes two forms of low-order iterative hard decoding + high-order iterative hard decoding and low-order iterative hard decoding + high-order iterative soft decoding, and the MLC soft iterative decoding process includes two forms of low-order soft decoding + high-order iterative soft decoding and low-order soft decoding + high-order iterative hard decoding.

And step 34, performing deinterleaving processing on the high-order iteration bit data and the low-order iteration bit data to obtain deinterleaved iteration data.

And step 35, performing RS decoding on the deinterleaved iterative data to obtain RS iterative decoding data. In this embodiment, a hard decoding method is adopted when RS decoding the deinterleaved iterative data.

And step 36, sending the RS iterative decoding data to the information sink.

It should be noted that, when the first demodulation process is performed on the channel received data transmitted by the channel in the hard decision manner to obtain the low-order demodulated data, an iterative decoding manner needs to be used, and the obtained decoded data can meet the performance requirement of the ethernet network with the data rate of 400Gb/s or more. That is, when the first demodulation process is performed in the hard decision manner, the flow of the concatenated decoding method includes MLC hard decoding + RS decoding + MLC hard iterative decoding + RS decoding.

When the first demodulation process is executed to the modulation data transmitted from the channel in a soft decision manner to obtain the low-order demodulation data, the method flow of the cascade decoding can include a non-iterative manner, namely MLC soft decoding + RS decoding; an iterative decoding mode can also be adopted, and the iterative decoding mode comprises two modes, wherein one mode is MLC soft decoding + RS decoding + MLC soft iterative decoding + RS decoding, and the other mode is MLC soft decoding + RS decoding + MLC hard iterative decoding + RS decoding.

A concatenated decoding method provided in a second embodiment of the present application is used for decoding modulated data obtained by using a concatenated coding method provided in a first embodiment of the present application, where the concatenated decoding method includes performing a first demodulation process on channel received data according to a preset modulation manner to obtain low-order demodulated data; decoding the low-order demodulation data to obtain low-order bit data; processing the channel receiving data by using low-order bit data to obtain high-order bit data; performing de-interleaving processing on the high-order bit data and the low-order bit data to obtain de-interleaved data; RS decoding is carried out on the de-interleaved data to obtain RS decoded data; and transmitting the RS decoded data to the information sink. The cascade decoding method uses the low-order bit data to assist in processing the high order in the modulation data to obtain the high-order bit data, thereby ensuring the decoding performance, reducing the complexity of the system and reducing the time delay.

In order to further embody the performance advantages of the cascade coding method and the cascade decoding method provided by the application, a PAM4 modulation mode is taken as an example, and a comparison experiment is performed.

In contrast to experiment 1, experiment group 1 uses the RS code + MLC code concatenated coding method provided in the first embodiment of the present application, specifically, the outer code uses RS (544, 514, t = 15), that is, KP4 FEC scheme, the inner code uses BCH (360, 320) code to encode the low-order bit data in the code composed of the MLC code, the error correction capability is 4, the high-order bit data is not encoded, there is no error correction capability, and the preset interleaving depth I =2. Among them, the interleaving depth I =2 has a great potential to be compatible with PCS (Physical Coding Sublayer) of the existing ethernet. Decoding by using the MLC soft decoding + RS decoding cascade decoding method provided in the second embodiment of the present application, specifically, in the MLC soft decoding process, demodulating channel received data by using a soft decision manner to obtain low-order demodulated data, and decoding the low-order demodulated data by using a Chase-II algorithm to obtain low-order bit data; wherein the number of the turning bits in the Chase-II algorithm is set to be 3; demodulating the channel receiving data in a hard decision mode by using low-order bit data as assistance, and directly outputting a demodulation result to obtain high-order bit data; and de-interleaving the high-order bit data and the low-order bit data to obtain de-interleaved data, and performing RS hard decoding on the de-interleaved data to obtain an RS decoding result.

The comparison group 1 uses a cascade coding method of RS codes and BCH codes; specifically, the outer code adopts an RS (544,514,t = 15) code, i.e., KP4 FEC scheme, and the inner code adopts a BCH (360,340) code. The code rates of the concatenated codes in the experimental group and the comparative group 1 are the same, and the gray code is used in the modulation process of the PAM4 in the comparative group 1. Comparative group 2 used the KP4 FEC scheme.

The decoding performance corresponding to the three groups of encoding methods in comparative experiment 1 is shown in fig. 10. Wherein experimental group 1 corresponds to 2RS + MLC _flip3 in FIG. 10, comparative group 1 corresponds to 2RS + BCH360 (t = 2) Gray code _ flip3 in FIG. 10, and comparative group 2 corresponds to RS (544, 514) in FIG. 10. As can be seen from fig. 10, when the output BER (Bit Error rate) index reaches 1E-15, the SNR (Signal-to-Noise Ratio) of the experimental group and the comparative group 1 relative to the comparative group 2 is improved by more than 1dB, and the decoding performance of the experimental group is improved by about 0.05dB compared with the comparative group 1. Wherein the flip in fig. 10 refers to the least reliable number of bits (flips) in BCH soft decoding.

Comparative experiment 2 verifies whether iterative decoding is required when MLC employs hard decoding. In the cascade coding method of the experimental group 2, the preset interleaving depth I =4, and the rest of the settings are the same as those of the cascade coding method of the experimental group 1 in the comparative experiment 1. The experimental group 2 uses a concatenated decoding method of MLC hard decoding + RS decoding. The cascade coding method of the experimental group 3 is the same as that of the experimental group 2, and decoding is performed by using an iterative decoding mode of MLC hard decoding, RS decoding, MLC hard iterative decoding and RS decoding.

The decoding performance corresponding to the two experimental groups in comparative experiment 2 is shown in fig. 11, where FER (Frame Error Rate) of the experimental group 2 corresponds to FER-sim 4kp4 iter1 in fig. 11, BER (bit Error Rate) of the experimental group 2 corresponds to BER-sim4kp4 iter1 in fig. 11, FER (Frame Error Rate) index of the experimental group 3 corresponds to FER-sim 4kp4 iter2 in fig. 11, and BER index of the experimental group 3 corresponds to BER-sim4kp4 iter2 in fig. 11. The performance requirement of the Ethernet with the speed of 400Gb/s is that the output BER is E-15 when the raw-BER is 2E-3. As can be seen from FIG. 11, the decoding scheme of experiment set 2 requires about 1.8E-3 channel condition to satisfy the output BER of E-15, which does not meet the performance requirement of Ethernet; the decoding scheme of experimental group 3 meets the requirements at a channel condition of about 2.8E-3, far exceeding the performance requirements of ethernet.

It can be seen from the comparative experiment 2 that when the first demodulation process is performed on the modulated data transmitted from the channel in the hard decision manner to obtain the low-order demodulated data, an iterative decoding manner is required, and the obtained decoded data can meet the performance requirement of the ethernet network with the data rate of 400Gb/s or more.

Comparative experiment 3 the decoding performance of experimental group 1, experimental group 4 and experimental group 5 and comparative group 1 were compared. The cascade coding method of the experimental group 4 is the same as that of the experimental group 1, and the cascade decoding method adopts an iterative decoding mode of MLC soft decoding + RS decoding + MLC soft decoding + RS decoding, wherein because the high-order bit data is not coded during coding, the high-order bit data is subjected to hard decision demodulation by using the assistance of a low-order decoding result in the MLC soft decoding process, and the high-order bit data is directly output. The cascade coding method of the experimental group 5 is the same as that of the experimental group 1, and the cascade decoding method adopts an iterative decoding mode of MLC soft decoding, RS decoding, MLC hard iterative decoding and RS decoding.

The decoding performance results of comparative experiment 3 are shown in fig. 12; wherein, comparative group 1 corresponds to 2rs + bch360 (t = 2) gray code _ flip3 in fig. 12, experimental group 1 corresponds to 2rs + mlc _flip3 in fig. 12, experimental group 4 corresponds to 2rs + mlc two soft iterations _ flip3 in fig. 12, and experimental group 5 corresponds to 2rs + mlc soft and hard iterations _ flip3 in fig. 12. As can be seen from fig. 12, when the interleaving depth is 2, the fitting extension of the decoding performance curve is performed, and when the output BER is equal to 1E 15, the decoding performance of the experimental group 4 is improved by about 0.28dB compared with the experimental group 1, and the decoding performance of the experimental group 5 is improved by about 0.12dB compared with the experimental group 1. From the above analysis, it can be seen that the decoding performance of the concatenated decoding method can be increased by means of iterative decoding.

The specific embodiments provided above are only a few examples of the general inventive concept and do not limit the scope of the claims. Any other embodiments extended according to the scheme of the present application without inventive efforts will be within the scope of protection of the present application for a person skilled in the art.

Claims

1. A method of concatenated coding, comprising:

receiving information source data sent by an information source;

carrying out multilayer coding on the bit data of a preset number of groups to obtain multilayer code word data; the multi-layer code word data comprises a preset number of groups of multi-layer sub-code word data, and the length of each group of multi-layer sub-code word data is consistent; the length of check bits in the multi-layer sub-code word data corresponding to the most significant bit in the preset modulation mode is less than the length of check bits in the multi-layer sub-code word data corresponding to the least significant bit in the preset modulation mode;

and modulating the multilayer code word data according to the preset modulation mode to obtain modulation data for channel transmission.

2. The concatenated coding method of claim 1, wherein when the predetermined modulation scheme is PAM4, the predetermined number of sets of bit data includes a set of high-order bit data and a set of low-order bit data;

the multi-layer coding is performed on the bit data of a preset number of groups to obtain multi-layer codeword data, and the multi-layer codeword data includes:

coding the low-order bit data according to a first code rate to obtain low-order multilayer sub-code data;

wherein the first code rate is less than the second code rate, the high-order multi-layer sub-codeword data corresponds to the most significant bits in the PAM4 modulation scheme, and the low-order multi-layer sub-codeword data corresponds to the least significant bits in the PAM4 modulation scheme.

3. The concatenated coding method of claim 1, wherein when the predetermined modulation scheme is PAM4, the predetermined number of sets of bit data includes a set of high-order bit data and a set of low-order bit data;

the multi-layer coding is performed on the bit data of the preset number of groups to obtain multi-layer code word data, and the multi-layer code word data comprises

the high-order multi-layer sub-code data corresponds to the most significant bit in the PAM4 modulation mode, and the low-order multi-layer sub-code data corresponds to the least significant bit in the PAM4 modulation mode.

4. A concatenated decoding method, wherein the concatenated decoding method is used for decoding the modulation data obtained by the concatenated coding method according to any one of claims 1-3; the cascade coding method comprises the following steps:

executing a first demodulation process on channel receiving data according to a preset modulation mode to obtain a first quantity of groups of low-order demodulation data;

decoding the low-order demodulation data to obtain low-order bit data;

using the low-order bit data to assist in processing the channel receiving data to obtain a second number of groups of high-order bit data; the first number plus the second number is equal to a preset number, and the preset number is the number of bits corresponding to one symbol in the preset modulation mode;

de-interleaving the high-order bit data and the low-order bit data to obtain de-interleaved data;

performing RS decoding on the deinterleaved data to obtain RS decoded data;

and sending the RS decoding data to a signal sink.

5. The concatenated decoding method of claim 4, wherein when the predetermined modulation scheme is PAM4, the first number and the second number are both equal to 1; the lower bit data corresponds to the least significant bit of the modulated data, and the upper bit data corresponds to the most significant bit of the modulated data;

the processing the channel received data with the assistance of the low-order bit data to obtain high-order bit data includes:

when the most significant bit in the modulation data is not coded in the multi-layer coding process, the low-order bit data is used for assisting to execute a second demodulation process on the most significant bit of the channel receiving data in a hard decision mode to obtain the high-order bit data;

when the most significant bit in the modulation data is coded in a multilayer coding process, the low-order bit data is used for assisting to execute a third demodulation process on the most significant bit of the channel receiving data in a hard decision mode or a soft decision mode to obtain high-order demodulation data, and the high-order demodulation data is decoded to obtain the high-order bit data; when a third demodulation process is executed by adopting hard decision, decoding the high-order demodulation data by adopting a hard decoding mode; and when the third demodulation process is executed in a soft decision mode, decoding the high-order demodulation data in a soft decoding mode.

6. The concatenated decoding method of claim 5, wherein before the RS-decoded data is transmitted to a sink, the concatenated decoding method comprises:

the RS iterative decoding data is sent to an information sink;

the iterative decoding is performed on the RS decoding data to obtain RS iterative decoding data, and the iterative decoding method comprises the following steps:

interweaving the RS decoding data to obtain iterative input data; wherein the iterative input data comprises lower iterative input data and higher iterative input data;

when the most significant bit in the modulation data is not coded in the multi-layer coding process, the low-order iteration bit data is used for assisting to execute a second demodulation process on the high-order iteration input data in a hard decision mode, and the high-order iteration bit data is obtained;

when the most significant bit in the modulation data is coded in a multilayer coding process, the low-order iteration bit data is used for assisting in executing a third demodulation process on the high-order iteration input data in a hard decision mode or a soft decision mode, and a demodulation result is decoded to obtain high-order iteration bit data; when the third demodulation process is executed in a hard decision mode, decoding the demodulation result in a hard decoding mode; when the third demodulation process is executed in a soft decision mode, decoding the demodulation result in a soft decoding mode;

performing RS decoding on the de-interleaved iterative data to obtain RS iterative decoding data;

and sending the RS iterative decoding data to a signal sink.

7. The concatenated decoding method of claim 6, wherein when the lower iterative input data is decoded by a hard decoding method, the check bits in the lower iterative input data are set as the check bits in the lower bit data.

8. The concatenated decoding method of claim 6, wherein when the low-order iterative input data is decoded by soft decoding, a maximum likelihood ratio of information bits in the low-order iterative data is set to be most reliable, and a maximum likelihood ratio of check bits in the low-order iterative input data is set to be a maximum likelihood ratio in the low-order demodulated data.

9. The concatenated decoding method of claim 7, wherein the low-order demodulated data is obtained by a hard decision method when the first demodulation process is performed on the channel received data according to a PAM4 modulation scheme;

and obtaining the low-order bit data by adopting a hard decoding mode when decoding the low-order demodulation data.

10. The concatenated decoding method of any one of claims 5, 7, and 8, wherein the low-order demodulated data is obtained by using a soft-decision method when performing a first demodulation process on the channel received data according to a PAM4 modulation method;

and when decoding the low-order demodulated data, a soft decoding mode is adopted to obtain the low-order bit data.