maki/openairinterface5g
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

fix(nrLDPC_coding_segment): move rate matching, interleaving and their counterparts to segment coding implementation

Browse Source
This commit is contained in:
Romain Beurdouche
2025-01-08 08:13:43 +00:00
parent f9bff3d6a0
commit 244355a3d2
11 changed files with 444 additions and 393 deletions
Download Patch File Download Diff File Expand all files Collapse all files
openair1/PHY/CODING/coding_defs.h
+70 -116
View File
@@ -41,7 +41,6 @@
#define MAX_LDPC_ITERATIONS_MBSFN 4
#define LTE_NULL 2
#define NR_NULL 2
typedef struct {
unsigned short nb_bits;
unsigned short f1;
@@ -60,10 +59,9 @@ typedef struct interleaver_codebook {
extern t_interleaver_codebook *f1f2mat;
extern short *il_tb;
/** @addtogroup _PHY_CODING_BLOCKS_
* @{
*/
*/
/** \fn lte_segmentation(uint8_t *input_buffer,
uint8_t **output_buffers,
@@ -95,8 +93,6 @@ int32_t lte_segmentation(uint8_t *input_buffer,
uint32_t *Kminus,
uint32_t *F);
/** \fn uint32_t sub_block_interleaving_turbo(uint32_t D, uint8_t *d,uint8_t *w)
\brief This is the subblock interleaving algorithm from 36-212 (Release 8, 8.6 2009-03), pages 15-16.
This function takes the d-sequence and generates the w-sequence. The nu-sequence from 36-212 is implicit.
@@ -105,18 +101,17 @@ This function takes the d-sequence and generates the w-sequence. The nu-sequenc
\param w Pointer to output (w-sequence, interleaver output)
\returns Interleaving matrix cardinality (\f$K_{\pi}\f$ from 36-212)
*/
uint32_t sub_block_interleaving_turbo(uint32_t D, uint8_t *d,uint8_t *w);
uint32_t sub_block_interleaving_turbo(uint32_t D, uint8_t *d, uint8_t *w);
/** \fn uint32_t sub_block_interleaving_cc(uint32_t D, uint8_t *d,uint8_t *w)
\brief This is the subblock interleaving algorithm for convolutionally coded blocks from 36-212 (Release 8, 8.6 2009-03), pages 15-16.
This function takes the d-sequence and generates the w-sequence. The nu-sequence from 36-212 is implicit.
\brief This is the subblock interleaving algorithm for convolutionally coded blocks from 36-212 (Release 8, 8.6 2009-03),
pages 15-16. This function takes the d-sequence and generates the w-sequence. The nu-sequence from 36-212 is implicit.
\param D Number of input bits
\param d Pointer to input (d-sequence, convolutional code output)
\param w Pointer to output (w-sequence, interleaver output)
\returns Interleaving matrix cardinality (\f$K_{\pi}\f$ from 36-212)
*/
uint32_t sub_block_interleaving_cc(uint32_t D, uint8_t *d,uint8_t *w);
uint32_t sub_block_interleaving_cc(uint32_t D, uint8_t *d, uint8_t *w);
/** \fn void sub_block_deinterleaving_turbo(uint32_t D, int16_t *d,int16_t *w)
\brief This is the subblock deinterleaving algorithm from 36-212 (Release 8, 8.6 2009-03), pages 15-16.
@@ -125,16 +120,16 @@ This function takes the w-sequence and generates the d-sequence. The nu-sequenc
\param d Pointer to output (d-sequence, turbo code output)
\param w Pointer to input (w-sequence, interleaver output)
*/
void sub_block_deinterleaving_turbo(uint32_t D, int16_t *d,int16_t *w);
void sub_block_deinterleaving_turbo(uint32_t D, int16_t *d, int16_t *w);
/** \fn void sub_block_deinterleaving_cc(uint32_t D, int8_t *d,int8_t *w)
\brief This is the subblock deinterleaving algorithm for convolutionally-coded data from 36-212 (Release 8, 8.6 2009-03), pages 15-16.
This function takes the w-sequence and generates the d-sequence. The nu-sequence from 36-212 is implicit.
\brief This is the subblock deinterleaving algorithm for convolutionally-coded data from 36-212 (Release 8, 8.6 2009-03),
pages 15-16. This function takes the w-sequence and generates the d-sequence. The nu-sequence from 36-212 is implicit.
\param D Number of input bits
\param d Pointer to output (d-sequence, turbo code output)
\param w Pointer to input (w-sequence, interleaver output)
*/
void sub_block_deinterleaving_cc(uint32_t D,int8_t *d,int8_t *w);
void sub_block_deinterleaving_cc(uint32_t D, int8_t *d, int8_t *w);
/** \fn generate_dummy_w(uint32_t D, uint8_t *w,uint8_t F)
\brief This function generates a dummy interleaved sequence (first row) for receiver, in order to identify
@@ -148,7 +143,8 @@ the NULL positions used to make the matrix complete.
uint32_t generate_dummy_w(uint32_t D, uint8_t *w, uint8_t F);
/** \fn generate_dummy_w_cc(uint32_t D, uint8_t *w)
\brief This function generates a dummy interleaved sequence (first row) for receiver (convolutionally-coded data), in order to identify the NULL positions used to make the matrix complete.
\brief This function generates a dummy interleaved sequence (first row) for receiver (convolutionally-coded data),
in order to identify the NULL positions used to make the matrix complete.
\param D Number of systematic bits plus 4 (plus 4 for termination)
\param w This is the dummy sequence (first row), it will contain zeros and at most 31 "LTE_NULL" values
\returns Interleaving matrix cardinality (\f$K_{\pi}\f$ from 36-212)
@@ -168,7 +164,8 @@ uint32_t generate_dummy_w_cc(uint32_t D, uint8_t *w);
uint8_t Nl,
uint8_t r)
\brief This is the LTE rate matching algorithm for Turbo-coded channels (e.g. DLSCH,ULSCH). It is taken directly from 36-212 (Rel 8 8.6, 2009-03), pages 16-18 )
\brief This is the LTE rate matching algorithm for Turbo-coded channels (e.g. DLSCH,ULSCH).
It is taken directly from 36-212 (Rel 8 8.6, 2009-03), pages 16-18 )
\param RTC R^TC_subblock from subblock interleaver (number of rows in interleaving matrix) for up to 8 segments
\param G This the number of coded transport bits allocated in sub-frame
\param w This is a pointer to the w-sequence (second interleaver output)
@@ -184,7 +181,6 @@ uint32_t generate_dummy_w_cc(uint32_t D, uint8_t *w);
\param nb_rb Number of PRBs
\returns \f$E\f$, the number of coded bits per segment */
uint32_t lte_rate_matching_turbo(uint32_t RTC,
uint32_t G,
uint8_t *w,
@@ -200,20 +196,19 @@ uint32_t lte_rate_matching_turbo(uint32_t RTC,
uint8_t nb_rb);
/**
\brief This is the LTE rate matching algorithm for Convolutionally-coded channels (e.g. BCH,DCI,UCI). It is taken directly from 36-212 (Rel 8 8.6, 2009-03), pages 16-18 )
\brief This is the LTE rate matching algorithm for Convolutionally-coded channels (e.g. BCH,DCI,UCI).
It is taken directly from 36-212 (Rel 8 8.6, 2009-03), pages 16-18 )
\param RCC R^CC_subblock from subblock interleaver (number of rows in interleaving matrix) for up to 8 segments
\param E Number of coded channel bits
\param w This is a pointer to the w-sequence (second interleaver output)
\param e This is a pointer to the e-sequence (rate matching output, channel input/output bits)
\returns \f$E\f$, the number of coded bits per segment */
uint32_t lte_rate_matching_cc(uint32_t RCC,
uint16_t E,
uint8_t *w,
uint8_t *e);
uint32_t lte_rate_matching_cc(uint32_t RCC, uint16_t E, uint8_t *w, uint8_t *e);
/**
\brief This is the LTE rate matching algorithm for Turbo-coded channels (e.g. DLSCH,ULSCH). It is taken directly from 36-212 (Rel 8 8.6, 2009-03), pages 16-18 )
\brief This is the LTE rate matching algorithm for Turbo-coded channels (e.g. DLSCH,ULSCH).
It is taken directly from 36-212 (Rel 8 8.6, 2009-03), pages 16-18 )
\param RTC R^TC_subblock from subblock interleaver (number of rows in interleaving matrix)
\param G This the number of coded transport bits allocated in sub-frame
\param w This is a pointer to the soft w-sequence (second interleaver output) with soft-combined outputs from successive HARQ rounds
@@ -265,7 +260,8 @@ uint32_t lte_rate_matching_turbo_rx_abs(uint32_t RTC,
uint32_t *E_out);
/**
\brief This is the LTE rate matching algorithm for Convolutionally-coded channels (e.g. BCH,DCI,UCI). It is taken directly from 36-212 (Rel 8 8.6, 2009-03), pages 16-18 )
\brief This is the LTE rate matching algorithm for Convolutionally-coded channels (e.g. BCH,DCI,UCI).
It is taken directly from 36-212 (Rel 8 8.6, 2009-03), pages 16-18 )
\param RCC R^CC_subblock from subblock interleaver (number of rows in interleaving matrix)
\param E This the number of coded bits allocated for channel
\param w This is a pointer to the soft w-sequence (second interleaver output) with soft-combined outputs from successive HARQ rounds
@@ -273,11 +269,7 @@ uint32_t lte_rate_matching_turbo_rx_abs(uint32_t RTC,
\param soft_input This is a pointer to the soft channel output
\returns \f$E\f$, the number of coded bits per segment
*/
void lte_rate_matching_cc_rx(uint32_t RCC,
uint16_t E,
int8_t *w,
uint8_t *dummy_w,
int8_t *soft_input);
void lte_rate_matching_cc_rx(uint32_t RCC, uint16_t E, int8_t *w, uint8_t *dummy_w, int8_t *soft_input);
/** \fn void ccodedot11_encode(uint32_t numbytes,uint8_t *inPtr,uint8_t *outPtr,uint8_t puncturing)
\brief This function implements a rate 1/2 constraint length 7 convolutional code.
@@ -286,10 +278,7 @@ void lte_rate_matching_cc_rx(uint32_t RCC,
@param outPtr Pointer to output buffer
@param puncturing Puncturing pattern (Not used at present, to be removed)
*/
void ccodedot11_encode (uint32_t numbytes,
uint8_t *inPtr,
uint8_t *outPtr,
uint8_t puncturing);
void ccodedot11_encode(uint32_t numbytes, uint8_t *inPtr, uint8_t *outPtr, uint8_t puncturing);
/*!\fn void ccodedot11_init(void)
\brief This function initializes the generator polynomials for an 802.11 convolutional code.*/
@@ -299,11 +288,6 @@ void ccodedot11_init(void);
\brief This function initializes the trellis structure for decoding an 802.11 convolutional code.*/
void ccodedot11_init_inv(void);
/** \fn void ccodelte_encode(int32_t numbits,uint8_t add_crc, uint8_t *inPtr,uint8_t *outPtr,uint16_t rnti)
\brief This function implements the LTE convolutional code of rate 1/3
with a constraint length of 7 bits. The inputs are bit packed in octets
@@ -314,12 +298,7 @@ void ccodedot11_init_inv(void);
@param outPtr Pointer to output buffer
@param rnti RNTI for CRC scrambling
*/
void
ccodelte_encode (int32_t numbits,
uint8_t add_crc,
uint8_t *inPtr,
uint8_t *outPtr,
uint16_t rnti);
void ccodelte_encode(int32_t numbits, uint8_t add_crc, uint8_t *inPtr, uint8_t *outPtr, uint16_t rnti);
/*!\fn void ccodelte_init(void)
\brief This function initializes the generator polynomials for an LTE convolutional code.*/
@@ -337,13 +316,9 @@ void ccodedab_init(void);
\brief This function initializes the trellis structure for decoding an DAB convolutional code (first 3 bits).*/
void ccodedab_init_inv(void);
/*!\fn void crcTableInit(void)
\brief This function initializes the different crc tables.*/
void crcTableInit (void);
void crcTableInit(void);
/*!\fn uint32_t crc24a(uint8_t *inPtr, int32_t bitlen)
\brief This computes a 24-bit crc ('a' variant for overall transport block)
@@ -351,7 +326,7 @@ based on 3GPP UMTS/LTE specifications.
@param inPtr Pointer to input byte stream
@param bitlen length of inputs in bits
*/
unsigned int crc24a (unsigned char * inptr, int bitlen);
unsigned int crc24a(unsigned char *inptr, int bitlen);
/*!\fn uint32_t crc24b(uint8_t *inPtr, int32_t bitlen)
\brief This computes a 24-bit crc ('b' variant for transport-block segments)
@@ -359,7 +334,7 @@ based on 3GPP UMTS/LTE specifications.
@param inPtr Pointer to input byte stream
@param bitlen length of inputs in bits
*/
uint32_t crc24b (uint8_t *inPtr, int32_t bitlen);
uint32_t crc24b(uint8_t *inPtr, int32_t bitlen);
/*!\fn uint32_t crc24c(uint8_t *inPtr, int32_t bitlen)
\brief This computes a 24-bit crc ('c' variant for transport-block segments)
@@ -367,68 +342,67 @@ based on 3GPP Rel 15 specifications.
@param inPtr Pointer to input byte stream
@param bitlen length of inputs in bits
*/
uint32_t crc24c (uint8_t *inPtr, int32_t bitlen);
uint32_t crc24c(uint8_t *inPtr, int32_t bitlen);
/*!\fn uint32_t crc16(uint8_t *inPtr, int32_t bitlen)
\brief This computes a 16-bit crc based on 3GPP UMTS specifications.
@param inPtr Pointer to input byte stream
@param bitlen length of inputs in bits*/
unsigned int crc16 (unsigned char * inptr, int bitlen);
unsigned int crc16(unsigned char *inptr, int bitlen);
/*!\fn uint32_t crc12(uint8_t *inPtr, int32_t bitlen)
\brief This computes a 12-bit crc based on 3GPP UMTS specifications.
@param inPtr Pointer to input byte stream
@param bitlen length of inputs in bits*/
unsigned int crc12 (unsigned char * inptr, int bitlen);
unsigned int crc12(unsigned char *inptr, int bitlen);
/*!\fn uint32_t crc12(uint8_t *inPtr, int32_t bitlen)
\brief This computes an 11-bit crc based on 3GPP NR specifications.
@param inPtr Pointer to input byte stream
@param bitlen length of inputs in bits*/
unsigned int crc11 (unsigned char * inptr, int bitlen);
unsigned int crc11(unsigned char *inptr, int bitlen);
/*!\fn uint32_t crc8(uint8_t *inPtr, int32_t bitlen)
\brief This computes a 8-bit crc based on 3GPP UMTS specifications.
@param inPtr Pointer to input byte stream
@param bitlen length of inputs in bits*/
unsigned int crc8 (unsigned char * inptr, int bitlen);
unsigned int crc8(unsigned char *inptr, int bitlen);
/*!\fn uint32_t crc8(uint8_t *inPtr, int32_t bitlen)
\brief This computes a 6-bit crc based on 3GPP NR specifications.
@param inPtr Pointer to input byte stream
@param bitlen length of inputs in bits*/
unsigned int crc6 (unsigned char * inptr, int bitlen);
unsigned int crc6(unsigned char *inptr, int bitlen);
int check_crc(uint8_t *decoded_bytes, uint32_t n, uint8_t crc_type);
/*!\fn void phy_viterbi_dot11_sse2(int8_t *y, uint8_t *decoded_bytes, uint16_t n,int offset,int traceback)
\brief This routine performs a SIMD optmized Viterbi decoder for the 802.11 64-state convolutional code. It can be
run in segments with final trace back after last segment.
\brief This routine performs a SIMD optmized Viterbi decoder for the 802.11 64-state convolutional code.
It can be run in segments with final trace back after last segment.
@param y Pointer to soft input (coded on 8-bits but should be limited to 4-bit precision to avoid overflow)
@param decoded_bytes Pointer to decoded output
@param n Length of input/trellis depth in bits for this run
@param offset offset in receive buffer for segment on which to operate
@param traceback flag to indicate that traceback should be performed*/
void phy_viterbi_dot11_sse2(int8_t *y,uint8_t *decoded_bytes,uint16_t n);
void phy_viterbi_dot11_sse2(int8_t *y, uint8_t *decoded_bytes, uint16_t n);
/*!\fn void phy_viterbi_lte_sse2(int8_t *y, uint8_t *decoded_bytes, uint16_t n)
\brief This routine performs a SIMD optmized Viterbi decoder for the LTE 64-state tail-biting convolutional code.
@param y Pointer to soft input (coded on 8-bits but should be limited to 4-bit precision to avoid overflow)
@param decoded_bytes Pointer to decoded output
@param n Length of input/trellis depth in bits*/
//void phy_viterbi_lte_sse2(int8_t *y,uint8_t *decoded_bytes,uint16_t n);
void phy_viterbi_lte_sse2(int8_t *y,uint8_t *decoded_bytes,uint16_t n);
// void phy_viterbi_lte_sse2(int8_t *y,uint8_t *decoded_bytes,uint16_t n);
void phy_viterbi_lte_sse2(int8_t *y, uint8_t *decoded_bytes, uint16_t n);
/*!\fn void phy_generate_viterbi_tables(void)
\brief This routine initializes metric tables for the optimized Viterbi decoder.
*/
void phy_generate_viterbi_tables( void );
void phy_generate_viterbi_tables(void);
/*!\fn void phy_generate_viterbi_tables_lte(void)
\brief This routine initializes metric tables for the optimized LTE Viterbi decoder.
*/
void phy_generate_viterbi_tables_lte( void );
void phy_generate_viterbi_tables_lte(void);
/*!\fn int32_t rate_matching(uint32_t N_coded,
uint32_t N_input,
@@ -443,63 +417,43 @@ void phy_generate_viterbi_tables_lte( void );
@param off Offset for seed
*/
int32_t rate_matching(uint32_t N_coded,
uint32_t N_input,
uint8_t *inPtr,
uint8_t N_bps,
uint32_t off);
int32_t rate_matching(uint32_t N_coded, uint32_t N_input, uint8_t *inPtr, uint8_t N_bps, uint32_t off);
int32_t rate_matching_lte(uint32_t N_coded,
uint32_t N_input,
uint8_t *inPtr,
uint32_t off);
int32_t rate_matching_lte(uint32_t N_coded, uint32_t N_input, uint8_t *inPtr, uint32_t off);
unsigned int crcbit (unsigned char * inputptr, int octetlen, unsigned int poly);
unsigned int crcbit(unsigned char *inputptr, int octetlen, unsigned int poly);
void phy_viterbi_dot11(int8_t *,uint8_t *,uint16_t);
void phy_viterbi_dot11(int8_t *, uint8_t *, uint16_t);
/**
* \brief perform segmentation of a Transport Block (TB) into segments
* \param input_buffer buffer containing the TB data to be segmented, can be NULL
* \param output_buffer buffer containing the TB data after segmentation, can be NULL
* \param B size of the TB in bits
* \param C pointer to write the number of segments
* \param K pointer to write the payload size per segment
* \param Zout pointer to write the lifting size per segment
* \param F pointer to write the number of filler bits per segment
*/
int32_t nr_segmentation(unsigned char *input_buffer,
unsigned char **output_buffers,
unsigned int B,
unsigned int *C,
unsigned int *K,
unsigned int *Zout,
unsigned int *F,
uint8_t BG);
unsigned char **output_buffers,
unsigned int B,
unsigned int *C,
unsigned int *K,
unsigned int *Zout,
unsigned int *F,
uint8_t BG);
void nr_interleaving_ldpc(uint32_t E, uint8_t Qm, uint8_t *e,uint8_t *f);
void nr_deinterleaving_ldpc(uint32_t E, uint8_t Qm, int16_t *e,int16_t *f);
int nr_get_R_ldpc_decoder(int rvidx,
int E,
int BG,
int Z,
int *llrLen,
int round);
int nr_rate_matching_ldpc(uint32_t Tbslbrm,
uint8_t BG,
uint16_t Z,
uint8_t *w,
uint8_t *e,
uint8_t C,
uint32_t F,
uint32_t Foffset,
uint8_t rvidx,
uint32_t E);
int nr_rate_matching_ldpc_rx(uint32_t Tbslbrm,
uint8_t BG,
uint16_t Z,
int16_t *w,
int16_t *soft_input,
uint8_t C,
uint8_t rvidx,
uint8_t clear,
uint32_t E,
uint32_t F,
uint32_t Foffset);
/**
* \brief compute parameter R for NR LDPC decoder
* \param rvidx redundancy version index
* \param E size of the code segment in bits
* \param BG LDPC base graph number
* \param Z lifting size
* \param llrlen pointer to the size in bits of input llrs
* \round HARQ round index
*/
int nr_get_R_ldpc_decoder(int rvidx, int E, int BG, int Z, int *llrLen, int round);
decoder_if_t phy_threegpplte_turbo_decoder;
decoder_if_t phy_threegpplte_turbo_decoder8;