WO2009062425A1 - A method for matching turbo code rate and for reading code word bit - Google Patents

Abstract

A method for matching Turbo code rate comprises the following steps: (a) sending a information packet to a Turbo code encoder with 1/r code rate, and generating a systematic bit flow and (r-1) checked bit flow; (b) making the systematic bit flow and the (r-1) checked bit flow encoded by the Turbo encoder pass through a respective sub-interleaver respectively, upon interleaved, forming a circular buffer with the systematic bit flow being provided in front of a circular buffer, and the checked bit flow being provided after the systematic bit flow by interlacing; (c)reading E code word bits required by the hybrid automatic repeat request HARQ transmission from the circular buffer area in turn each time, and forming a HARQ sub-package. The orthogonal re-transmission of the Turbo code is realized and the Turbo decoding capability is optimized by the implement of the rate-matching algorithm based on a circular buffer provided by present invention. And the rate-matching algorithm provided by present invention does not need to define superfluous version number and can save the signaling cost.

Description

 Turbo code rate matching and code word bit reading method

Technical field

 The present invention relates to digital communication systems, and more particularly to a method for turbo code rate matching and codeword bit reading in channel coding of a digital communication system.

Background technique

 The digital communication system usually includes a source, a source encoder, a channel encoder and a modulator, and the receiving end usually includes a demodulator, a channel decoder, a source decoder and a sink, as shown in FIG. Show. The channel coder is used to introduce information bits into the redundant information according to certain rules so that the receiving channel decoder can correct the error occurring when the information is transmitted on the channel to some extent.

Among many channel coding techniques, Turbo codes are currently recognized as one of the best forward error correction codes. The error correction performance of Turbo codes is far superior to that of other codes, and the more decoding iterations, the better the performance of decoding and error correction. Therefore, it is often recommended to use in data transmission where reliability is very high. . For example, the third generation mobile communication uses the 8-state 1/3 bit rate binary turbo code as the channel coding standard. The commonly used binary Turbo coding is a parallel concatenated code with an internal interleaver, which is generally formed by parallel concatenation of two recursive systematic convolutional code (RSC) component code encoders of the same structure. The Turbo intra-code interleaver randomly replaces the bit positions in the input binary information sequence before the second component code encoder. When the interleaver is sufficiently large, the Turbo code has the characteristics of an approximately random long code. Such a binary Turbo code is used in WCDMA (Wideband Code Division Multiple Access) and TD-SCDMA (Time Division Synchronous Code Division Multiple Access), and the structure is as shown in FIG. The input binary information sequence; ^ generates a one-way 3 full sequence Z _k through the first component code encoder. At the same time, the input binary information sequence Xji is interleaved by the Turbo code interleaver, and the second component code encoder generates another check sequence Z _k '. At this time, if the coded bits are not punctured, the output code rate of the Turbo code is 1/3, and the coded bit sequence obtained by the output is: , ^, ^, ''^^ _1⁄2 , where is the length of the input binary information sequence. After all the information bit sequences have been encoded, the tail bits need to be taken out of the shift register feedback to perform the trellis operation termination. The first three tail bits are used to terminate the first component code encoder, and the last three tail bits are used to terminate the second component code encoder. According to the above operation, you can get 12 grids. The transmitted bit terminated by the shape operation, the bit order is:

 After adding the encoded bit sequence, a Turbo encoding is completed.

 In a typical digital communication system, when designing a coded modulation scheme, different order modulation modes (such as quadrature phase shift keying QPSK, 16QAM (Quadrature Amplitude Modulation), 64QAM, etc.) are usually set. Different codes (such as convolutional codes, convolutional Turbo codes, etc.), each code usually has a different code rate (Rate, such as 1/2, 2/3, 3/4, 5/6, etc.). When the system is scheduled, a specific coding modulation mode is arranged for each burst according to the channel quality and service requirements. In order to achieve better link adaptation, each code is better able to achieve a smaller granularity when transforming the code rate.

 For the Turbo code commonly used in digital communication systems, the increase of the code rate is obtained by puncturing the low code rate mother code to obtain higher code rate coding. We also classify this method as the rate. Match ( Rate Matching, or RM). For 3GPP GPP Turbo codes, the system also supports various possible code rates and hybrid automatic repeat request (HARQ) procedures through rate matching.

 As an alternative to the 3GPP Rel-6 rate matching algorithm, rate matching based on circular buffer

(Circular Buffer Rate Matching, CB RM) provides a method for simply generating a punctual pattern with excellent performance. The specific structure is shown in Figure 3. In the cyclic buffer rate matching method, each data stream will be rearranged by a respective sub-interleaver, which is called sub-block interleaver; usually, in order to simplify the hardware implementation, the number of columns of the block interleaver is fixed. The number of rows changes as the length of the interleave changes, so the circular buffer can be thought of as a row buffer of "R rows x C columns", that is, as a "R row x C column" virtual buffer. The intra-block interleaving used in the cyclic buffer matching method of 3GPP is a block interleaver with a fixed number of columns of 32 columns. Since there are three data streams of system bits, first parity bits and second parity bits in the loop buffer, the "loop buffer" can be regarded as a virtual buffer with 96 columns.

Then, in a single output buffer, the rearranged systematic bits are placed at the start position, and then two rearranged parity bit data streams are interleaved, referred to as inter-block interleaving. For the desired rate (rate), the cyclic buffer rate matching bit selection is to sequentially read L bits from the beginning of the buffer somewhere as a rate matched output. In general, the bits selected for transmission can be read from any point in the buffer, and if they reach the end of the buffer, they can be wound into the buffer. The starting position continues to read the data until the completion of reading L bits. In view of the convenience of hardware implementation, the bits selected for transmission are preferably read from a certain column start position of the virtual buffer instead of any one bit position.

In a HARQ system based on cyclic buffer rate matching, different redundancy versions (Xrv) can be specified by defining different starting points. For example, in the 3GPP system, the HARQ process based on the cyclic buffer rate matching defines four RV versions (RV = 0, 1, 2, 3). Each time the HARQ retransmits the L-long sub-packet, it starts from the starting point defined by the redundancy version and selects L bits clockwise.

 The introduction of the redundancy version helps to simplify the synchronous HARQ operation. However, the introduction of the redundancy version may cause the overlapping of the codewords corresponding to different HARQ sub-packets, and the redundancy version is also selected during the asynchronous HARQ operation. It needs to be controlled by signaling, thus increasing the signaling overhead of the system.

Summary of the invention

 The technical problem to be solved by the present invention is to provide a cyclic buffer rate matching method and a bit reading method, so that the HARQ of the turbo code is retransmitted to the optimized orthogonal retransmission; and the HARQ retransmission is not required to be defined. The redundancy version number can save signaling overhead.

 In order to solve the above technical problem, the present invention provides a method for rate matching of Turbo codes, including the following steps:

 (a) sending the information packet to a turbo code encoder having a code rate of 1/r, generating a system bit stream and (r-1) a checked bit stream;

 (b) The system bit stream programmed by the Turbo encoder and the (r-1) check bit stream are respectively passed through respective sub-interleavers, and after interleaving, the system bit stream is placed in front of the circular buffer. The bit stream is interleaved after the system bit stream to form a circular buffer;

 (c) sequentially reading E codeword bits required for each hybrid automatic repeat request HARQ transmission from the circular buffer to form a HARQ sub-packet.

Further, the foregoing method may further have the following feature, in the step (c), reading the HARQ When the required E codeword bits are transmitted, the reading begins from the next column of the column of the transmitted codeword bits of the previous HARQ transmission.

 Further, the above method may further have the following feature: the reading from the next column of the column of the codeword bits transmitted from the previous HARQ transmission refers to reading from the starting position of the next column.

 Further, the foregoing method may also have the following features: The reading position of each HARQ sub-package is determined according to the following principles:

 Let the length of the first n HARQ sub-packets be E i=0 , 1 ... n-1, then the current HARQ sub-packets are read from the ^: column, where the total of the virtual buffers is represented. The number of columns; ife/to is the number of offset columns in the first transmission; it indicates that the number of columns corresponding to the "R row x C column" virtual buffer is transmitted in the previous n transmissions.

 Further, the above method may further have the following features, and the C„ is further determined by:

C _n = ∑ ", indicating rounding up, and R is the number of rows in the virtual buffer.

 Further, the above method may also have the following characteristics: the bit position of the current n + 1th start reading is read according to the following principles:

K = R ((C _n + delta)%C _m

 Where R is the number of rows in the virtual buffer; the total number of columns in the virtual buffer; delta is the number of offset columns in the first transmission; indicating that the previous n transmissions have a total of "R rows XC columns" "The number of columns in the virtual buffer."

 Further, the above method may further have the following feature: if the codeword bit is read, the end of the circular buffer is reached, and the code sub-bit is continuously read from the beginning position of the circular buffer until the reading E is completed. Codeword bits.

 The present invention also proposes a method for reading codeword bits, which sequentially reads E codeword bits required for each HARQ transmission from a circular buffer, wherein the current nth + 1th start reading bit position Read according to the following principles:

K = R ((C _n + delta)%C _m

Where R is the number of rows in the virtual buffer; the total number of columns in the virtual buffer; delta is the number of offset columns in the first transmission; indicating that the previous n transmissions are transmitted in a corresponding "R" Row XC column" The number of columns in the virtual buffer.

 The cyclic buffer-based rate matching algorithm proposed by the present invention completely implements orthogonal retransmission of Turbo codes to optimize Turbo decoding performance; and the rate matching algorithm proposed by the present invention does not need to define a redundancy version number. Save signaling overhead.

BRIEF abstract

 1 is a schematic structural diagram of a digital communication system;

 Figure 2 is the structure of a Turbo encoder;

 Figure 3 is a structure of the existing cyclic buffer rate matching;

 Figure 4 is a structure of continuous transmission loop buffer rate matching;

 Figure 5 is a continuous transfer structure of virtual loop buffer rate matching.

Preferred embodiment of the invention

 In order to facilitate a deep understanding of the technical content of the present invention, the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

 The present invention is directed to the characteristics of the cyclic buffer rate matching, and proposes that the codeword bits selected for each HARQ retransmission sub-packet are immediately following the codeword bits of the previous HARQ sub-packet. Specifically, since the loop buffer can be regarded as a "virtual loop buffer" of "R row x C column", as shown in Fig. 5. For example, the circular buffer in the 3GPP rate matching algorithm can be thought of as a 96-column virtual buffer. If the previous HARQ sub-packet is transmitted to the i-th column, even if the column has not been transmitted yet, the current HARQ sub-packet reads the codeword bit from the i+1th column, and if it reaches the end of the buffer, The data is read until the start position of the buffer continues until the L bits are read, as shown in FIG.

 The present invention provides a method for Turbo code rate matching, comprising the following steps:

(a) The information packet is sent to a turbo code encoder having a code rate of 1/r, and a system bit stream and (r-1) parity bit streams are generated. When the code rate is 1/3, the check bit stream is two. When the code rate is 1/5, the check bit stream is

4;

 (b) The system bit stream programmed by the Turbo encoder and the (r-1) check bit stream are respectively passed through respective sub-interleavers, and after interleaving, the system bit stream is placed in front of the circular buffer. The bit stream is interleaved after the system bit stream to form a circular buffer;

 (c) sequentially reading E codeword bits required for each HARQ transmission from the circular buffer to form a HARQ sub-packet, wherein each time the circular buffer reads the codeword bits to form a HARQ sub-packet , the location of the read is from the next column of the column that the previous HARQ transmission has transmitted. For example, if the previous HARQ sub-packet is transmitted to the i-th column, even if the column has not been transmitted yet, the current HARQ sub-packet reads the codeword bit from the i+1th column.

 The technical content of the present invention will be further described below by taking the 1/3 code rate of 3GPP as an example. The present invention provides a Turbo code rate matching method, which includes the following steps:

 (a) sending a packet of length K to a 3GPP 1/3 bit rate turbo code encoder to generate a systematic bit stream and a first and second parity bit stream, since 12 tail bits are added, So each bit stream has a length of K+4.

(b) performing cyclic buffer-based rate matching on the codewords encoded by the Turbo encoder, and the system bit stream and the first and second parity bit streams are respectively passed through a sub-interleaver, where the sub-interleaver takes The number of columns in the 3GPP Turbo rate matching algorithm is 32 columns of sub-interleavers, so the number of rows of the sub-interleaver, that is, the length of each column is

 K + 4

 , here ", means rounding up,

 32 The system bit stream is then placed in front of the circular buffer, and the bit streams of the first and second parity are placed erroneously behind the system bit stream, resulting in a total of 96 columns of virtual circular buffers.

(c) sequentially reading the E codeword bits required for each HARQ transmission from the circular buffer to form a HARQ sub-packet. In particular, the reading position of each HARQ sub-packet is determined by the following process: Let the length of the first n HARQ sub-packets be =0, 1 ... n-1, first transmission or first transmission Its index i=0, the second transmission i=l, and so on, then the current n+1th transmission, i=n, starts reading the bit position: K = R ((C _n + delta)%C _m

Here, R is the number of rows of the virtual buffer (that is, the number of rows of the sub-interleaver); represents the total number of columns of the virtual buffer; ife/to is the number of offset columns at the time of the first transmission, and % indicates Modular operation. Therefore, the current HARQ sub-package is read from the (C _n + i^ column).

Which needs to be further determined by the following formula:

 i "," means rounding up, indicating that the previous n transfers have transmitted a total of the number of columns corresponding to the "R row X C column" virtual buffer.

Specifically, for the 3GPP cyclic buffer rate matching algorithm, it is equal to 96; the offset column number delta of the first transmission is 2 .

 So if the current transmission is the first time, that is, the first transmission, n=0, then it can be calculated as:

 C. =∑ = 0

R Therefore, the current HARQ sub-packets are read from the loop buffer's ^<3⁄4 ₊ ^%9 (5 positions), that is, E _Q bits are read from the ^ = R*2 position to form a HARQ sub-packet.

 If the current transmission is the second time, there is a previous transmission, ie n=l, then it can be calculated as:

So when the position starts reading

£ bits, forming a HARQ sub-packet.

If the current transmission is the third time, there are two transmissions in front, that is, n=2, then it can be calculated as:

 So the current HARQ sub-packet from the R of the circular buffer ( +2 %9 (5 position open)

E ₂ bits are read to form a HARQ sub-packet. And so on.

The invention may, of course, be embodied in a variety of other embodiments, and various modifications and changes can be made thereto in accordance with the present invention. The scope of protection.

Industrial applicability

 The cyclic buffer-based rate matching algorithm proposed by the present invention completely implements orthogonal retransmission of Turbo codes to optimize Turbo decoding performance; and the rate matching algorithm proposed by the present invention does not need to define a redundancy version number. Save signaling overhead.

Claims

Claim  1. A Turbo code rate matching method, comprising the following steps:  (a) sending the information packet to a turbo code encoder having a code rate of 1/r, generating a system bit stream and (r-1) a checked bit stream;  (b) The system bit stream programmed by the Turbo encoder and the (r-1) check bit stream are respectively passed through respective sub-interleavers, and after interleaving, the system bit stream is placed in front of the circular buffer. The bit stream is interleaved after the system bit stream to form a circular buffer;  (c) sequentially reading E codeword bits required for each hybrid automatic repeat request HARQ transmission from the circular buffer to form a HARQ sub-packet.  2. The method according to claim 1, wherein in the step (c), when the E codeword bits required for the HARQ transmission are read, the column of the transmitted codeword bits is transmitted from the previous HARQ transmission. The next column starts reading.  3. The method of claim 2, wherein the reading from the next column of the column of the previously transmitted HARQ transmitted codeword bits means reading from the beginning of the next column.  4. The method according to claim 1 or 2 or 3, wherein: the reading position of each HARQ sub-package is determined according to the following principles:  Let the length of the first n HARQ sub-packets be E i=0 , 1 ... n-1, then the current HARQ sub-packets are read from the ^: column, where the total of the virtual buffers is represented. The number of columns; ife/to is the number of offset columns in the first transmission; it indicates that the number of columns corresponding to the "R row x C column" virtual buffer is transmitted in the previous n transmissions.  The method according to claim 4, wherein: the further determining is: c„= ", indicating an up-rounding operation, and R represents a number of rows of the virtual buffer.

6. The method according to claim 5, wherein: the bit position of the current n+1th start reading is read according to the following principle: K = R ((C _n + delta)%C _m Where R represents the number of rows of the virtual buffer; represents the total number of columns of the virtual buffer; delta Is the number of offset columns at the time of the first transfer; indicates that the number of columns of the virtual buffer corresponding to the "R row XC column" has been transmitted for the previous n transfers.  7. The method according to claim 1 or 2 or 3, wherein, if the codeword bit is read, reaching the end of the circular buffer, reading continues from the beginning of the circular buffer The code bits are bits until the E codeword bits are read.  8. A method for reading codeword bits, characterized by: sequentially reading E codeword bits required for each HARQ transmission from a circular buffer, wherein the current nth + 1th start reading The bit position is read according to the following principles: K = R ((C _n + delta)%C _m  Where R is the number of rows in the virtual buffer; the total number of columns in the virtual buffer; delta is the number of offset columns in the first transmission; indicating that the previous n transmissions have a total of "R rows XC columns" "The number of columns in the virtual buffer."