WO2018082286A1

WO2018082286A1 - Encoding and decoding method and apparatus for serial communication system based on serdes technology

Info

Publication number: WO2018082286A1
Application number: PCT/CN2017/085860
Authority: WO
Inventors: 徐小明; 康婕; 赵羽; 项圣文
Original assignee: 深圳国人通信股份有限公司
Priority date: 2016-11-04
Filing date: 2017-05-25
Publication date: 2018-05-11
Also published as: CN106411322A

Abstract

Provided are an encoding and decoding method and apparatus for a serial communication system based on SerDes technology. The method comprises an encoding step: dividing 18-bit common data into two pieces of 9-bit original data; respectively using the two pieces of original data as the initial address N of two code look-up tables, and converting the initial addresses N into 2*N+0 and 2*N+1 as the input address of the code look-up table; outputting a first code word and a second code word comprising imbalance information according to the input addresses, the number of '1' in the first code word being larger than or equal to the number of '0', and the number of '0' in the second code word being larger than or equal to the number of '1'; outputting a selection indication signal according to the motion polarity difference value of a current code stream; and selecting a code word from the first code word and second code word output by each look-up table according to the selection indication signal to form a next output code word. The decoding process of the scheme is the inverse process of encoding, and the implementation method is similar to the encoding process. The present invention achieves the target of direct current balance of the output code stream, and at the same time, improves the transmission performance.

Description

Instruction manual

Title of Invention: Codec Method and Apparatus Based on SerDes Technology Serial Communication System

Technical field

[0001] The present invention relates to the field of communications technologies, and in particular, to a code decoding method and apparatus based on a SerDes technology serial communication system.

Background technique

[0002] With the continuous development of communication technologies, communication systems have increasingly increased transmission bandwidth requirements, and high-speed SerD es (serializer/deserializer) serial communication technology has high channel capacity, low system cost, and fast transmission speed. And so on, this makes SerDes-based serial communication technology gradually become the mainstream high-speed data communication technology.

[0003] The structure of the serial communication system based on SerDes is as shown in FIG. 1: On the transmitting end, the parallel data is encoded by the primary encoder 1; the parallel-serial converter 2 receives the encoded parallel data, according to the clock manager 4 The provided cuckoo clock signal converts the parallel data into serial data and transmits the serial data via the transmitter 3. At the receiving end, the receiver 7 receives the serial input data and sends it to the serial to parallel converter 6, which converts the serial input data into parallel data and sends it to the main decoder 5, the main The decoder 5 decodes it and outputs it; the clock manager 4 recovers the chirp signal from the data through the serial-to-parallel converter.

[0004] One of the key technologies in the SerDes-based serial communication system is the chirp and data recovery at the receiving end, and the main factors affecting the receiving end clock and data recovery performance are the level conversion and DC balance of the serial data stream. . Level shifting refers to the conversion of data from "0" to "1" or from "1" to "0". DC balance means that the number of "0" and "1" in the serial data stream is basically the same.

[0005] In order to ensure that the serial data stream has good level shifting and DC balance characteristics, the serial communication system based on SerDes technology encodes the input parallel data at the transmitting end.

[0006] After performing the encoding process, the main considerations include:

[0007] (1), code word imbalance D (Disparity), represents the difference between the number of 1 in a code word and the number of 0. Positive quantity (Positive Disparity) indicates that the number of 1 in the code word is greater than 0, and the "N" in the positive N polarity indicates that the number of 1 in the data is more than the number of 0; Negative Disparity Table The number of 1s in the code word is less than the number of 0, and the "N" in the negative N polarity indicates that the number of 1s in the data is less than the number of 0's. For example: For 0100, D=-2; For 0111, D = +2.

[0008] (2) Run the polarity difference value RD (Running Disparity). Indicates the difference between the number of 1s and 0s in all currently consecutive codewords, that is, the imbalance of the current code stream. The total imbalance of the output code stream is represented by the running polarity difference value and the RDS (Running Disparity Sum) value. In coding, consider the "earth" polarity of the RDS of the output stream, and do not consider the specific values. "+" indicates that the number of "1"s in the current stream is greater than the number of "0"; "-" indicates that the number of "0"s in the current stream is more than the number of "1"s. For a good line code, R DS should be a finite value with an absolute value close to zero.

[0009] (3) Code Run Length Indicates the number of consecutive 1s or 0s in a continuous serial code stream.

[0010] In the prior art, the most common line coding technique is 8B/10B coding, and 8B/10B coding will be 8Bit (8 bits).

The data is encoded into lOBit (10-bit) data. Its good DC balance and short run length (maximum run length is only 5) make it easy for the receiver to recover data and clocks. However, the 8B/10B code also has obvious disadvantages, that is, the coding efficiency is not high. If the transmission rate is a useful data stream of lGb/s, the actual line transmission rate after encoding is 1.25 Gb/s, which is nearly 25% redundant. Such large redundant credits will become unacceptable in higher-rate optical communication systems (transmission rates of up to 6Gb/s or even lOGb/s); in addition, as the performance of fiber-optic communication equipment increases, the optical communication system pairs the lines. The code requirements are much reduced. For example, the maximum coded run length is less than 20 and can be easily detected by the receiver.

technical problem

Problem solution

Technical solution

[0011] The features and advantages of the invention are set forth in part in the description which follows,

[0012] In order to overcome the problems of the prior art, the present invention provides a codec method and apparatus based on the SerDes technology serial communication system, which adopts a 9B/10B code and uses a lookup table method in an FPGA (Field-Programmable Gate Array, field). Implemented within a programmable gate array) to increase efficient transmission efficiency.

[0013] The technical solution adopted by the present invention to solve the above technical problems is as follows:

[0014] According to an aspect of the present invention, a codec side in a serial communication system based on SerDes technology is provided Method, including coding steps:

[0015] dividing 18-bit ordinary data into two 9-bit original data;

[0016] The two original data are respectively used as the initial address N of the two encoded lookup tables;

[0017] converting the initial address into 2*N+0 and 2*N+1 as input addresses of the code lookup table;

[0018] the code lookup table outputs a first codeword and a second codeword including unbalance information according to the input address; wherein, the number of "1"s in the first codeword is greater than or equal to "0" The number of "unbalances is positive; the number of "0"s in the second codeword is greater than or equal to the number of ", and the degree of imbalance is negative;

[0019] outputting a selection indication signal according to a running polarity difference value of the current code stream;

[0020] selecting one codeword among the first codeword and the second codeword outputted by each of the lookup tables according to the selection indication signal to form a next output codeword.

[0021] Optionally, the first codeword and the second codeword are obtained by adding 10-bit codeword information and 3-bit unbalance degree D coding information; wherein the 10-bit codeword is The information is a mapping of the original data of 9 bits to 10 bits; the degree of imbalance D = n - (10-n), where n is the number of "1"s in the code word.

[0022] Optionally, the K code encoding step is further included:

[0023] defining six K codes, each of which is a group of K codes, and the code words in each group of K codes are complementary to each other;

Receiving an input K code indication signal and a polarity selection signal;

[0025] K code encoding consisting of two 10 bits is output through the K code lookup table.

Optionally, the selection indication signal is output according to the running polarity difference value of the current code stream: if the running polarity difference value of the current code stream is negative, the selection indication of the output The signal is high level; if the running polarity difference value of the current code stream is positive, the output selection indication signal is low level

[0027] selecting one codeword in the first codeword and the second codeword outputted in each of the lookup tables according to the selection indication signal to form an output codeword 吋: if the output selection indication signal is high Ping, in the next output codeword, selecting the first codeword in the first codeword and the second codeword; if the output selection indication signal is low, the next The second codeword is selected in the first codeword and the second codeword in the output codeword.

[0028] Optionally, the calculating the running polarity difference value of the current code stream specifically includes the following steps:

And [0029] the algebraic sum of the running polarity difference value and the unbalanced degree of the output codeword as the encoded polarity Total difference value;

[0030] when the output codeword is perfectly balanced, the running polarity difference value of the current code stream remains unchanged; [0031] when the output codeword is unbalanced:

[0032] if the sum of the polarity differences after the encoding is less than 0, the running polarity difference value of the current code stream is negative [0033], if the sum of the polarity differences after the encoding is greater than 0, then The running polarity difference value of the current code stream is positive

[0034] if the sum of the polarity differences after the encoding is equal to 0, and the running polarity difference value of the original code stream is positive, the running polarity difference value of the current code stream is negative;

[0035] If the sum of the polarity differences after encoding is equal to 0, and the running polarity difference value of the original code stream is negative, the running polarity difference value of the current code stream is positive.

[0036] Optionally, the step of decoding is further included:

[0037] processing the decoded output data at the corresponding common data position in the decoding lookup table;

[0038] processing the decoded output data at the corresponding K code position in the decoding lookup table;

[0039] in the above generated decoding lookup table, respectively, according to the address of the lookup table to select a decoded output data;

[0040] The lower 9 bits of the two decoded output data are integrated and output.

[0041] According to another aspect of the present invention, a codec device in a serial communication system based on SerDes technology is provided, comprising: a main encoder, a line polarity calculator connected to the main encoder, An encoding selector coupled to the primary encoder;

[0042] The primary encoder includes:

[0043] a splitting unit, configured to divide 18-bit ordinary data into two original data of a high 9-bit and a low 9-bit; [0044] an initial address defining unit, configured to use two of the original data as two The initial address N of the code lookup table;

[0045] an input address calculation unit, configured to convert the initial address into 2*N+0 and 2*N+1 as an input address of the code lookup table;

[0046] an output unit, configured to output, according to the input address, a first codeword and a second codeword including unbalance information; wherein, the number of "1"s in the first codeword is greater than or equal to "0 The number of "unbalances is positive; the number of "0"s in the second codeword is greater than or equal to the number of ", and the degree of imbalance is negative; [0047] the line polarity calculator is configured to output a selection indication signal according to the running polarity difference value of the current code stream; [0048] the code selector is configured to output in each of the lookup tables according to the selection indication signal Selecting one of the first codeword and the second codeword constitutes an output codeword.

[0049] Optionally, further comprising a K code encoder connected to the line polarity calculator: the line polarity calculator is further configured to output a polarity selection signal;

[0050] The K code encoder includes:

[0051] a defining unit, configured to define six K codes, each of which is a group of K codes, and the code words in each group of K codes are complementary to each other;

[0052] a receiving unit, configured to receive the input K code indication signal and the polarity selection signal;

[0053] The K code encoding output unit is configured to output K code code composed of two 10 bits through the K code lookup table.

[0054] Optionally, the line polarity calculator includes:

[0055] a difference sum value calculation unit, configured to use an algebraic sum of the running polarity difference value and the unbalanced degree of the output codeword as a coded polarity difference sum value;

[0056] a balance determining unit, configured to determine whether the output codeword is perfectly balanced;

[0057] a running polarity difference value calculating unit, configured to: when the output codeword is perfectly balanced, the running polarity difference value of the current code stream remains unchanged;

[0058] The running polarity difference value calculating unit is further configured to: when the output codeword is unbalanced: if the encoded polarity difference sum value is less than 0, the running pole of the current code stream The difference value of the polarity difference is negative; if the sum of the polarity differences after the encoding is less than 0, the running polarity difference value of the current code stream is positive; if the sum of the polarity differences after the encoding is equal to 0, and If the running polarity difference value of the original code stream is positive, the running polarity difference value of the current code stream is negative; if the encoded polarity difference sum value is equal to 0, and the running polarity difference of the original code stream If the value is negative, the running polarity difference value of the current code stream is positive.

[0059] Optionally, further comprising a main decoder and a data integrator connected to the main decoder;

[0060] the main decoder is configured to process the decoded output data in the corresponding common data position in the decoding lookup table.

[0061] processing the decoded output data at the corresponding K code position in the decoding lookup table;

[0062] selecting a decoded output data according to the address of the lookup table in the generated decoding lookup table;

[0063] The data integrator is configured to integrate the lower 9 bits of the two decoded output data and output. Advantageous effects of the invention

Beneficial effect

The present invention provides a codec method and apparatus based on a SerDes technology serial communication system, which maps 9-bit original words into 10-bit code words. The encoded data stream has a good DC balance characteristic with a maximum run length of 7. These advantages make the data transmission performance on the link greatly improved, and the chirp recovery is also easy at the receiving end, which is very suitable for high-speed serialization. The transmission of data. Compared to the currently used 8B/10B encoding method, the effective transmission rate of the 9B/10B encoding scheme is increased by 12.5% at the same transmission rate.

[0065] Those skilled in the art will better understand the features and contents of these technical solutions by reading the specification.

Brief description of the drawing

DRAWINGS

The present invention will be more apparent from the following description of the embodiments of the present invention. Any limit in the sense, in the drawing:

1 is a block diagram showing the structure of a SerDes communication system in the prior art.

2 is a schematic flow chart of an encoding step in a serial communication system based on SerDes technology according to an embodiment of the present invention;

3 is a diagram showing a polarity deviation state transition according to an embodiment of the present invention;

4 is a schematic flow chart of a K code encoding step in a serial communication system based on SerDes technology according to an embodiment of the present invention;

5 is a schematic flow chart of a decoding step in a serial communication system based on SerDes technology according to an embodiment of the present invention;

6 is a schematic structural diagram of a serial communication coding system based on SerDes technology according to an embodiment of the present invention;

7 is a schematic structural diagram of a main encoder in a serial communication system based on SerDes technology according to an embodiment of the present invention;

8 is a schematic structural diagram of a line polarity calculator in a serial communication system based on SerDes technology according to an embodiment of the present invention;

9 is a schematic structural diagram of a serial communication system based on SerDes technology according to another embodiment of the present invention; 10 is a schematic structural diagram of a K code encoder in a serial communication system based on SerDes technology according to an embodiment of the present invention;

11 is a schematic structural diagram of a serial communication system based on SerDes technology according to still another embodiment of the present invention;

12 is a schematic diagram showing an output input relationship of an implementation process of K code encoding.

Embodiments of the invention

[0079] The present invention can map a 9-bit original word into a 10-bit code word by designing a code lookup table. The code lookup table can be implemented as follows.

[0080] First, considering all 10-bit codeword samples, a 10-bit codeword sample has 2 ^Λ 10=1024 combinations. The 1024 codeword vectors are grouped according to the number of bits "1" to obtain the distribution pattern shown in Table 1 below:

[] [Table 1]

Table 1 Classification of 10-bit codewords

[0082] In all possible 1024 codeword samples, all codeword samples whose codeword imbalance D has an absolute value greater than 4 are discarded, so that 252 perfectly balanced codewords (D=0) and 330 pairs are left. Perfectly balanced codewords (D=±2, ±4). That is to say, there are 582 pairs of code words, more than 9 pairs of 512 pairs of arbitrary original words.

[0083] To minimize the maximum run length, 70 pairs of unneeded codewords are picked out in 582 pairs of codewords. The principle is to consider the head and tail run length of each codeword, for example, the codeword 0010111111 has a head run length of 2 and a tail run length of 6. The above 582 pairs of codewords are all analyzed in this way, and the distribution law shown in Table 2 can be obtained: [] [Table 2]

Table 2 codeword run length distribution table

[0085] 32 pairs of codewords with head and tail run lengths greater than 5 (where codewords 0000011111 and 1111100000 are counted twice in the table). Then there are 550 pairs of codewords left, and the length of the head and tail runs is

4, so the maximum run of the continuous stream is 8.

[0086] The reason that all headers with a head and tail run length of 4 cannot be discarded is that too many codewords are lost, resulting in less than 512 pairs of codewords required for 9-bit encoding. Here only the codewords with a tail run length of 4 are discarded.

[0087] The last available codeword obtained by the above analysis is 232 balanced codewords, the run of the head and the end is not more than _3; 200x2 codewords of D=±2, 80x2 D=±4 The codeword has a run of no more than 4 at the head and end, and 232+200+80=512, which just satisfies the 512 pairs of codeword samples required for 9-bit encoding.

[0088] After obtaining 512 pairs of codeword samples required for 9-bit encoding, a mapping of 9 bits to 10 bits may be performed, including the following steps:

[0089] Step 1: First, the first bit of the 232 balanced code words is removed, and the remaining 9 bits are the original words corresponding thereto. For example, if the balance code word is 0100100111, the first digit is 100100111, 100100111 is the original word corresponding to it, that is, it is 295 coded or 295 in decimal, and the first code word and the second code word of these original words are the same. [0090] Step 2: The remaining codewords are analyzed with the original words, and some of the unbalanced codewords may be associated with the original words of the sections. The rule is to process the codeword with D=±2 first, post processing

D=±4 codeword. For example, the 010011000 original word can correspond to 1010011000 or 0010011000, because the code word of D=±2 is processed first, so it will be

1010011000 corresponds. After this processing, there are 228 original words to find the coding result of one mode in the first code word and the second code word.

[0091] Step 3: The remaining 52 original words cannot be encoded in a certain mode by the above method because the imbalance is too large. In this case, only the remaining codewords can be found in the remaining allowed codewords. For example, if the original word 101000000, the closest one among the remaining codeword sets is 0101000001, then 0101000001 is the second codeword of 101000000.

[0092] Step 4: After the above steps are processed, each 9-bit original word finds a corresponding codeword mode, and the codeword mode may be the first codeword or the second codeword. In addition to balancing the codewords in step 1, the other 280 original words require another mode of encoding. Therefore, you need to find the codeword in the other mode of the 280 original words. The coding criterion is: For a original word that needs to be encoded, it is only necessary to find the nearest one from the remaining code words, and the polarity of the code word that has been found by steps 1~3 is opposite. For example, the original word 000000000 finds that the second code word is 0001000011 according to step 3, and its polarity is negative; then the first code word of the original word should select one of the remaining allowed code words that is closest to him and has a positive polarity. Codeword, here select 1001010111 as its first codeword.

[0093] Step 5: The mapping of 9 bits to 10 bits can be realized by the above steps, but the coding table is not only composed of 10-bit code word information, but also contains the imbalance D information of each code word. The specific operation method is: first calculate the number n of "1" in each codeword, then the codeword unbalance degree D=n-(10-n), since the codeword is filtered, so n is only possible here. It is 3, 4, 5, 6, and 7. Therefore, the code word imbalance D can only be -4, -2, 0, +2, +4. The codeword unbalance D is 3-bit encoded as shown in Table 3 below:

[] [Table 3] Codeword imbalance -4 -2 0 2 4

Degree D value

Unbalance D 110 111 000 001 010 Value corresponding code Table 3 code word imbalance D value coding table

[0095] The coded information of the codeword unbalance degree D in the table is shifted to the left by 10 and the codeword information is added to obtain the final code, as shown in Equation 1: In the formula, F_Code is the final coded information in the code lookup table, and D_Code represents the codeword. The encoding information of the unbalance degree D, U_Code indicates the codeword information corresponding to the 9-bit original word. Multiplying D_Code by 2 to the 10th power means shifting 10 bits to the left. Therefore, the final encoded information is a 13-bit encoded data. The 512 9-bit original words are mapped to the first code word and the second code word of RD- and RD+, respectively, so there are 1024 code words in the code table. Each codeword is processed according to Equation 1 to generate a final code table and stored in the lookup table inside the FPGA.

[0096] F_Code = D_Code*2 ^A 10 + U_Code (Equation 1)

[0097] Based on the implementation method of the above code lookup table, a serial communication system codec method and apparatus based on SerDes technology of the present invention is provided.

As shown in FIG. 2, the present invention provides a codec method in a serial communication system based on SerDes technology, which includes a data 9B/10B encoding step:

[0099] Sl, the 18-bit ordinary data is divided into two raw data of high 9 bits and low 9 bits;

[0100] S2, the two original data are respectively used as the initial address of the two code lookup tables N (N=0~51 l)

[0101] S3, converting the initial address into 2*N+0 and 2*N+1 as an input address of the code lookup table; [0102] S4. The code lookup table is output according to the input address. a first codeword and a second codeword of the imbalance information;

[0103] In the present invention, in order to ensure the DC balance of the output code stream, two 10-bit code words are represented by the code lookup table mapping, and a 9-bit original word is respectively used as the first code word and the second code word; , the number of "1" in the first codeword is greater than or equal to the number of "0", and the number of "0" in the second codeword is greater than or equal to the number of ". If the 10-bit codeword is perfectly balanced, ie If the number of "1" in the codeword is the same as the number of "0", then the first codeword and the second codeword are the same; if the 10-bit codeword is not perfectly balanced, that is, the number of "1" in the codeword is If the number of "0" is different, the first codeword and the second codeword are two different codewords of opposite polarity;

[0104] The first codeword and the second codeword are obtained by adding the codeword information of 10 bits and the coding information of the unbalance degree D of 3 bits; wherein, the 10-bit codeword information is 9 bits. The mapping of the original data to 10 bits; the imbalance degree 1)=^ (10-n), where _n is the number of "1"s in the codeword; [0105] S5. Output a selection indication signal according to a running polarity difference (RD) value of the current code stream;

[0106] Specifically, the selection indication signal is output according to the running polarity difference value of the current code stream: if the running polarity difference value of the current code stream is negative, the output selection indication signal is a high level; If the running polarity difference value of the current code stream is positive, the output selection indication signal is low level;

[0107] S6, selecting one codeword among the first codeword and the second codeword outputted by each of the lookup tables according to the selection indication signal to form a next output codeword;

[0108] Specifically, if the output selection indication signal is a high level, the first codeword is selected in the first codeword and the second codeword in the next output codeword; The selection indication signal is low, and the second codeword is selected in the first codeword and the second codeword in the next output codeword.

[0109] In an embodiment of the present invention, the running polarity difference (RD) value of the current code stream in step S5 is calculated according to the polarity difference sum (RDS) value indicating the encoded total unbalance degree information. The rules are shown in Figure 3:

[0110] The RD value of the initial engraving of the power-on is negative, and the RDS value is 0;

[0111] an algebraic sum of the imbalance between the original RD value and the output codeword is taken as the encoded RDS value;

[0112] When the output codeword is perfectly balanced, the RD value of the current code stream remains unchanged;

[0113] When the output codeword is unbalanced:

[0114] If the encoded RDS value is less than 0, the RD value of the current code stream is negative;

[0115] If the encoded RDS value is greater than 0, the RD value of the current code stream is positive;

[0116] If the encoded RDS value is equal to 0, and the RD value of the original code stream is positive, the RD value of the current code stream is negative; [0117] If the encoded RDS value is equal to 0, and the original code stream is RD If the value is negative, the RD value of the current stream is positive.

[0118] As shown in FIG. 4, in addition to the encoding step, a K code encoding step is also included:

[0119] Sl l, defining six K codes, each of the two K codes is a group, and the code words in each group of K codes are complementary to each other;

[0120] The K code can be used to perform functions such as byte alignment, frame delimitation, character padding, and clock recovery of the code stream.

In this embodiment, the 9B/10B coding code defines three groups of six K codes, which are K001 and K101, Κ002 and K102, and Κ003 and K103, respectively. Two codewords in each group of weights complement each other, respectively. The first codeword and the second codeword of the weight are encoded and output;

[0121] S12. Receive an input K code indication signal and a polarity selection signal.

[0122] S13. Output a K code encoded by two 10-bits through a K code lookup table. [0123] The implementation process of K code encoding is also equivalent to a lookup table, and its output input relationship is as shown in FIG.

[0124] As shown in FIG. 5, in another embodiment of the present invention, a decoding step is further included, where the decoding step and the encoding step are the same, and are also implemented based on a lookup table method, and the difference is that the translation is performed. The code process is memoryless. The so-called memoryless decoding refers to decoding a currently received 10-bit codeword to obtain a 9-bit data, which has nothing to do with the previous code stream, that is, only after receiving the same 10-bit codeword decoding. There will be a decoding result. Specifically, the decoding step may include:

[0125] S21. Perform processing on the decoded output data in the corresponding common data position in the decoding lookup table.

[0126] Decoding is the inverse of encoding, and is specifically implemented based on a lookup table method. As known from the above encoding process, in order to ensure the DC balance of the output code stream, two 10-bit code words are used in the code table mapping to represent a 9-bit original word, which are respectively the first code word and the second code word. If the 10-bit codeword is perfectly balanced, the first codeword and the second codeword are the same; if the 10-bit codeword is not perfectly balanced, the first codeword and the second codeword are different.

[0127] When the decoding lookup table is generated, if the first codeword and the second codeword are the same, only one location in the decoding lookup table stores the original word information. If the first codeword and the second codeword are different, there are two locations in the decoding lookup table to store the original word information. For example, the original word 000000000 (decimal is 0) is encoded to obtain two different code words, which are the first codeword 1001010111 (decimal is 599) and the second codeword 0001000011 (decimal is 67). Then, after decoding, the lookup table addresses are 599 and 67吋, and the decoded output will output 0, that is, the original word information is stored in both positions in the decoding table. The original word 0011100001 (decimal is 225) is encoded to obtain two identical codewords, that is, the first codeword and the second codeword are 1011100001 (decimal is 737), then after decoding, the lookup table address has Only equal to 737 吋, the decoding will output 225, that is, only one position in the decoding table stores 225 original word information.

[0128] S22. Process the decoded output data at the corresponding K code position in the decoding lookup table.

[0129] Specifically, the decoded output data corresponding to the K code position includes the lower 9 bits of the K code original word information and the upper 3 bits are the K code indication information;

[0130] In the decoding lookup table generation process, K code indication information should also be included. The invention is shared during the encoding process

The SJ3 group has six K codes, K001 and K101, Κ002 and Κ102, and Κ003 and Κ103, respectively. Two codewords in each set of codes are complementary to each other, respectively representing the first codeword and the second code of the _κ code. Word output. After generating the decoding lookup table, when K001 and K101吋 are detected, the first bit of the weight indication information is set to 1; when Κ002 and K102吋 are detected, the second bit of the weight indication information is set to 1; when Κ003 and K103吋, weight indication The third bit of the information is set to 1. Then, the 3-bit K code indication information is combined with the K code original word information, and the processing manner is as shown in Equation 2:

F_data = K_Flag[9]*2 ^A 9 + K_Flag[10]*2 ^A 10+ K_Flag[l 1]*2 ^Λ 11+K_data (Equation 2) [0132] where F_data decodes the K code in the lookup table The decoded output data, K_Flag is a 3-digit weight indication information, and K_data is a 9-digit weight original word information. After processing by Equation 2, the 1st, 2nd, and 3rd bits of the 3-bit K code indication information are corresponding to the 9th, 10th, and 11th bits of the decoded data output, and the 9-bit K-code original word information is combined, and the final translation is performed. The code output F_data is a 12-bit data.

[0133] It is worth mentioning that the decoding output of the normal data in the decoding lookup table is actually 12 bits. It also includes 3-bit K code indication information, just as normal data, and the 3 bits corresponding to the K code indication information are all 0.

[0134] S23. Select one decoded output data according to the address of the lookup table in the generated decoding lookup table.

[0135] The 20-bit decoded data is divided into two addresses of the upper 10 bits and the lower 10 bits as the addresses of the two decoding lookup tables; since the decoding process is memoryless, the two decoding lookup tables are based on The input address selects the corresponding decoded output data.

[0136] S24. Integrate the lower 9 bits of the two decoded output data and output.

[0137] That is, the two-way lookup table decoded output data is integrated into one 18-bit data and used as a decoded output. Its input and output relationship is:

[0138] Xout[17:0]={DoutB[8:0], DoutA[8:0] }

Where Xout is the decoded output signal, DoutA[8 _: 0] is the lower 9 bits of the lower 10 bit codeword lookup table output decoding data; DoutB[8 _: 0] is the upper 10 bit codeword lookup table The lower 9 bits of the decoded data are output. In the decoded output data stream described in S24, the SerDes system does not know which is the K code which is normal data, so the decoding process also needs to output a K code indication signal, that is, Kout[l _: 0], when it is When two bits are 0吋, the system will consider the current decoded output to be the decoded output of normal data. When it is not 0吋, it indicates that the current decoded output is the decoded output of the K code. The K-code indication signal plays a key role in restoring the 吋 clock for the SerDes system to maintain link synchronization. Specifically, the K code information (ie, the 13th to 15th bits of the two lookup table output data) data acquired in step S22 is input, and the K code indication signal is output through the related logical operation. Its input and output relationship is: [0140] When DoutB[10]=l吋, Kout[l]=l;

[0141] When DoutA[9]=l吋, Kout[0]=0; DoutA[l l]=l吋, Kout[0]=l.

[0142] wherein Kout is a K code indication signal, DoutA is a low 10-bit codeword lookup table outputting decoded data; DoutB is a high 10-bit codeword lookup table outputting decoded data.

The encoding and decoding method in the serial communication system based on the SerDes technology in the embodiment of the present invention is described above, and the following is a serial communication system based on the SerDes (serializer/deserializer) technology in the embodiment of the present invention. The codec device is described.

[0144] As shown in FIG. 6, the present invention further provides a codec device in a serial communication system based on SerDes technology, comprising a main encoder 10, a line polarity calculator 20 connected to the main encoder, and a main encoder A connected code selector 30.

[0145] As shown in FIG. 7, wherein the main encoder 10 includes:

[0146] The splitting unit 11 is configured to divide the normal data of 18 bits into two original data of the upper 9 bits and the lower 9 bits; [0147] an initial address defining unit 12, configured to use two original data as two The initial address of the code lookup table N (N=0~51 l);

[0148] The input address calculation unit 13 is configured to convert the initial address into 2*N+0 and 2*N+1 as the code lookup table input address;

[0149] The output unit 14 is configured to output, according to the input address, the first codeword and the second codeword including the imbalance information according to the input address. In the present invention, in order to ensure the DC balance of the output codestream, the mapping in the code lookup table is performed. 2Using two 10-bit codewords to represent a 9-bit original word, respectively a first codeword and a second codeword; wherein, the number of "1"s in the first codeword is greater than or equal to the number of "0" The number of "0" in the second codeword is greater than or equal to the number of ". If the 10-bit codeword is perfectly balanced, the first codeword and the second codeword are the same; if the 10-bit codeword is not perfectly balanced, then The first codeword and the second codeword are different and the polarities are opposite; the first codeword in the above code lookup table and the codeword information in which the second codeword is 10 bits are added to the coded information of the unbalanced degree D of 3 bits. The obtained 10-bit codeword information is 9-bit original data corresponding to a 10-bit coding mapping; wherein the imbalance degree D=n-( 10-n ), where n is the number of "1"s in the codeword .

[0150] The line polarity calculator 20 described above is used to ensure DC balance of the output code stream, which provides a selection indication signal for the code selector. In the encoding process, the output of the selection indication signal is determined based on the running polarity difference RD value indicating the current code stream imbalance characteristic. Specifically, the value is output according to the running polarity difference value of the current code stream. The selection indication signal 吋: if the running polarity difference value of the current code stream is negative, the output selection indication signal is high level; if the running polarity difference value of the current code stream is positive, the output selection indication signal Is low.

[0151] The function of the code selector 30 is to select a coded output, which is composed of three inputs and one output, which are respectively a data input terminal 8, a data input terminal B, an input selection indication signal, and a data output terminal. Its output and input relationship is:

[0152] 1) when the selection indication signal is 1吋, the code output = input data A;

[0153] 2) When the selection indication signal is 0吋, the code output = input data ^

[0154] In another embodiment of the present invention, referring to FIG. 8, the line polarity calculator 20 includes:

[0155] The difference sum value calculation unit 21 is configured to use the algebraic sum of the running polarity difference value and the unbalanced degree of the output code word as the encoded polarity difference sum value;

[0156] The balance determining unit 22 is configured to determine whether the output codeword is perfectly balanced;

[0157] The running polarity difference value calculating unit 23 is configured to: when the output codeword is perfectly balanced, the running polarity difference value of the current code stream remains unchanged;

[0158] The running polarity difference value calculating unit 23 is further configured to: when the output code word is unbalanced: if the encoded polarity difference sum value is less than 0, the current code stream running polarity difference value is negative; If the sum of the polarity differences after encoding is greater than 0, the running polarity difference value of the current code stream is positive; if the sum of the polarity differences after encoding is equal to 0, and the running polarity difference value of the original code stream is positive, then The running polarity difference value of the current code stream is negative; if the encoded polarity difference sum value is equal to 0, and the running polarity difference value of the original code stream is negative, the running polarity difference value of the current code stream is positive. It should be noted that the running polarity difference value at the initial power-on encapsulation is negative, and the sum of the polarity differences after encoding is 0.

[0159] The code selector 30 is configured to select one codeword among the first codeword and the second codeword outputted by each lookup table according to the selection indication signal to form a next output codeword.

As shown in FIG. 9 and FIG. 10, the K code encoder 40 is further included in the embodiment, and is connected to the line polarity calculator 20 and the code selector 30. The K code encoder 40 includes:

[0161] The defining unit 41 is configured to define six K codes, each of the two K codes is a group, and the code words in each group of K codes are complementary to each other; the K code can be used to complete the byte alignment of the code stream. , frame delimitation, character padding, and clock recovery, etc. In this embodiment, the 9B/10B codec defines 3 groups of 6 K codes, which are K001 and K101, Κ002 and K102, and K003 and K103, two codewords in each set of weights are complementary to each other, respectively representing the first codeword and the second codeword coded output of the weight;

[0162] The receiving unit 42 is configured to receive the input weight indication signal and the polarity selection signal;

[0163] The weight code output unit 43 is configured to output a weight code composed of two 10 bits through a weight lookup table;

The implementation process of the weight code is also equivalent to a lookup table. The output and input relationship is shown in Table 4.

[0164] Although not shown in the figure, it further includes a delay module connected to the weight encoder 40. The function of the delay module is to ensure that the encoder code output and the weight output indication signal are aligned, that is, the encoder outputs the indication signal at the weight. Effectively, the encoded output of this 吋 encoder is exactly the weight. In order to ensure this alignment relationship, the weight input indication signal must obtain the weight output indication signal through the delay module, and the delay of the delay module must be equal to the delay of the input data after the encoder encoding process.

[0165] As shown in FIG. 10, in this embodiment, a main decoder 50 and a data integrator 60 connected to the main decoder are further included;

[0166] The main decoder 50 is configured to process the decoded output data in the corresponding common data position in the decoding lookup table; and process the decoded output data in the corresponding weight position in the decoding lookup table; In the above-mentioned generated decoding lookup table, one decoding output data is respectively selected according to the address of the lookup table; decoding is the inverse process of encoding, and is specifically implemented based on the lookup table method. As can be seen from the above, in order to ensure the DC balance of the output code stream, two 10-bit code words are used in the code table mapping to represent a 9-bit original word, which are the first code word and the second code word, respectively. If the 10-bit codeword is perfectly balanced, the first codeword and the second codeword are the same; if the 10-bit codeword is not perfectly balanced, the first codeword and the second codeword are not identical. Therefore, when the decoding lookup table is generated, if the first codeword and the second codeword are the same, only one location in the decoding lookup table stores the original word information. If the first code word and the second code word are different, there are two locations in the decoding lookup table to store the original word information. For example, the original word 000 000000 (decimal 0) is encoded to obtain two different code words, which are the first code word 1001010111 (decimal is 599) and the second code word 0001000011 (decimal is 67). Then, after decoding, the lookup table addresses are 599 and 67吋, and the decoded output will output 0. The original word 0011100001 (decimal is 225) is encoded to obtain two identical codewords, that is, RD- and the second codeword are 1011100001 (decimal is 737), then after decoding, the lookup table address has and only equals 737吋, decoding only outputs 225.

[0167] The main decoder 50 can be used for decoding the weight; specifically, the special processing is performed on the original information of the weight in the generation of the decoding table, that is, the original word information of the 9-digit weight is added to the 3 Bit weight indication information. The invention is encoded In the process, 3 groups of 6 K codes are shared, which are K001 and Κ101, Κ002 and Κ102, and Κ003 and K10 respectively.

3. The two code words in each set of weights are complementary to each other, and represent the first code word and the second code word code output of the weight respectively. After generating the decoding table, when K001 and K101吋 are detected, the first bit of the weight indication information is set to 1; when Κ002 and K102吋 are detected, the second bit of the weight indication information is set to 1; when Κ003 and K103 are detected Inches

The third bit of the weight indication information is set to 1.

[0168] The data integrator 60 is configured to integrate the lower 9 bits of the two decoded output data and output the data; specifically, combine the decoded data of the two lookup tables into one 18-bit data, and use it as the whole main Decoded output of the decoder. Its input and output relationship is:

[0169] Xout[17:0]={DoutB[8:0], DoutA[8:0] }

[0170] wherein Xout is a decoded output signal, and DoutA is a low 10-bit codeword lookup table to output decoded data; Dout

B outputs decoded data for a high 10-bit codeword lookup table.

[0171] In addition, the K code indication information processor 70 is further configured to input the acquired data including the K code information (ie, the 9th to 11th bits of the two lookup table output data), and obtain the K through the related logical operation. The code indicates the signal output. Its input and output relationship is:

[0172] When DoutB[10]=l吋, Kout[l]=l;

[0173] When DoutA[9]=l吋, Kout[0]=0; DoutA[l l]=l吋, Kout[0]=l.

[0174] wherein Kout is a K code indication signal, DoutA is a low 10-bit codeword lookup table outputting decoded data; DoutB is a high 10-bit codeword lookup table outputting decoded data.

[0175] The 9B/10B encoding according to the present invention is mainly implemented in a FPGA by a lookup table method, which divides ordinary data into high 9 bits and low 9 bits respectively as addresses of a lookup table, and a lookup table puts addresses. The corresponding positive polarity (Rd+) first codeword and negative polarity (Rd-) second codeword are output to the code selector, and the code selector selects the corresponding coded output through the indication signal provided by the line polarity calculator. . The K code encoder selects the corresponding K code output according to the K code indication and the indication provided by the line polarity calculator, and the final code output is selected by the code selector in the K code output and the normal data code output.

The preferred embodiments of the present invention have been described above with reference to the drawings, and the present invention can be implemented by various modifications without departing from the scope and spirit of the invention. For example, features illustrated or described as part of one embodiment can be used in another embodiment to yield a further embodiment. The above is only a preferred embodiment of the present invention, and is not intended to limit the scope of the present invention. Equivalent variations made by the contents of the drawings are included in the scope of the invention.

Claims

Claim

[Claim 1] A codec method in a serial communication system based on SerDes technology, comprising an encoding step: dividing 18-bit ordinary data into two original data of 9 bits and 9 bits;

Taking the two original data as the initial address N of the two encoded lookup tables respectively;

Converting the initial address into 2*N+0 and 2*N+1 as input addresses of the code lookup table;

The code lookup table outputs a first codeword and a second codeword including unbalance information according to the input address; wherein, the number of "1"s in the first codeword is greater than or equal to "0" Number, the degree of imbalance is positive; the number of "0" in the second codeword is greater than or equal to the number of "1"

, its imbalance is negative;

Outputting a selection indication signal according to a running polarity difference value of the current code stream;

And selecting one codeword among the first codeword and the second codeword outputted by each of the lookup tables according to the selection indication signal to form a next output codeword.

[Claim 2] The codec method in the serial communication system based on the SerDes technology according to claim 1, wherein the first codeword and the second codeword are 10-bit codeword information and 3-bit The coding information of the unbalance degree D is added; wherein, the 10-bit codeword information is a mapping of the original data of 9 bits to 10 bits; the imbalance degree D=n-(10-n) , where n is the number of "1"s in the codeword.

[Claim 3] The codec method in the serial communication system based on the SerDes technology according to claim 1, further comprising a K code encoding step:

Defining 6 K codes, each of the two K codes is a group, and the code words in each group of K codes are complementary to each other; receiving the input K code indication signal and the polarity selection signal;

The K code code consisting of two 10 bits is output through the K code lookup table.

[Claim 4] A codec method in a serial communication system based on SerDes technology according to claim 1, wherein

Outputting the selection indication signal according to the running polarity difference value of the current code stream: if the running polarity difference value of the current code stream is negative, the output selection indication signal is a high level; The running polarity difference value of the current code stream is positive, and the selection of the output is The indication signal is low;

Selecting one codeword in the first codeword and the second codeword outputted in each of the lookup tables according to the selection indication signal to form an output codeword 吋: if the output selection indication signal is high level, Selecting the first codeword in the first codeword and the second codeword in the next output codeword; if the output selection indication signal is a low level, the next output codeword Selecting the second codeword in the first codeword and the second codeword.

[Claim 5] A codec method in a serial communication system based on SerDes technology according to claim 1 or 4.

The calculation of the running polarity difference value of the current code stream specifically includes the step of combining the algebraic sum of the running polarity difference value and the output code word imbalance as the sum of the encoded polarity differences Value

When the output codeword is perfectly balanced, the running polarity difference value of the current code stream remains unchanged.

When the output codeword is unbalanced:

If the sum of the polarity differences after the encoding is less than 0, the running polarity difference of the current code stream is negative;

If the sum of the polarity differences after the encoding is greater than 0, the running polarity difference of the current code stream is positive;

If the sum of the polarity differences after the encoding is equal to 0, and the running polarity difference value of the original code stream is positive, the running polarity difference value of the current code stream is negative;

If the sum of the polarity differences after the encoding is equal to 0, and the running polarity difference value of the original code stream is negative, the running polarity difference value of the current code stream is positive.

[Claim 6] The codec method in the serial communication system based on the SerDes technology according to claim 1, further comprising a decoding step:

Processing the decoded output data at the corresponding normal data position in the decoding lookup table; processing the decoded output data at the corresponding K code position in the decoding lookup table; and performing the decoding lookup table generated above Selecting a decoded output data according to the address of the lookup table; The lower 9 bits of the two decoded output data are integrated and output.

[Claim 7] A codec device in a serial communication system based on a SerDes technology, comprising: a main encoder, a line polarity calculator connected to the main encoder, and the main encoder Code selector

The primary encoder includes:

Split unit, which is used to divide 18-bit ordinary data into two original data of 9 bits and 9 bits;

An initial address defining unit, configured to use two pieces of the original data as initial addresses N of two code lookup tables;

An input address calculation unit, configured to convert the initial address into 2*N+0 and 2*N+1 as an input address of the code lookup table;

An output unit, configured to output, according to the input address, a first codeword and a second codeword including unbalance information; wherein, the number of "1"s in the first codeword is greater than or equal to "0" a number whose imbalance is positive; wherein the number of "0" in the second codeword is greater than or equal to the number of "unbalances";

The line polarity calculator is configured to output a selection indication signal according to a running polarity difference value of the current code stream;

The code selector is configured to select one codeword among the first codeword and the second codeword outputted by each of the lookup tables according to the selection indication signal to form a next output codeword.

[Claim 8] The codec device in the serial communication system based on the SerDes technology according to claim 7, further comprising a K code encoder connected to the line polarity calculator: the line polarity The calculator is also used to output a polarity selection signal;

The K code encoder includes:

a defining unit for defining six K codes, each of which is a group of K codes, and the code words in each group of K codes are complementary to each other;

a receiving unit, configured to receive an input K code indication signal and a polarity selection signal;

The K code encoding output unit is configured to output a K code code composed of two 10 bits through a K code lookup table.

[Claim 9] The codec device in the serial communication system based on the SerDes technology according to claim 7, wherein the line polarity calculator comprises:

a difference sum value calculation unit, configured to use an algebraic sum of the unbalanced degree of the running polarity difference value and the output codeword as a sum of polarity differences after encoding; a balance determining unit, configured to determine the output code Whether the word is perfectly balanced;

a running polarity difference value calculating unit, configured to: when the output codeword is perfectly balanced, the running polarity difference value of the current code stream remains unchanged;

The running polarity difference value calculating unit is further configured to: when the output codeword is unbalanced: if the encoded polarity difference sum value is less than 0, the current code stream running polarity difference value If the sum of the polarity differences after the encoding is greater than 0, the running polarity difference value of the current code stream is positive; if the encoded polarity difference sum value is equal to 0, and the original code stream The running polarity difference value is positive, then the current polarity difference value of the current code stream is negative; if the encoded polarity difference sum value is equal to 0, and the original code stream running polarity difference value is negative Then, the running polarity difference value of the current code stream is positive.

[Claim 10] The codec device in the serial communication system based on the SerDes technology according to claim 7, further comprising a main decoder and a data integrator connected to the main decoder The decoder is configured to process the decoded output data in the corresponding common data position in the decoding lookup table; and select a decoded output data according to the address of the lookup table in the generated decoding lookup table;

The data integrator is configured to integrate the lower 9 bits of the two decoded output data and output