Skip to content

O
OpenXG-RAN
Commit c35025f5 authored by Quency Lin's avatar Quency Lin
Browse files
Options

Feat/ Implement decode of transform precoding, QPSK for now 

Refactor, write a new inner_rx()
Performance, Do unscrambling in thread
parent 420b5f13
Hide whitespace changes
Inline Side-by-side
Showing with 159 additions and 24 deletions
openair1/PHY/NR_TRANSPORT/nr_ulsch_demodulation.c
View file @ c35025f5
...@@ -1881,6 +1881,105 @@ void inner_rx_64qam_2layer (NR_DL_FRAME_PARMS *frame_parms, ...@@ -1881,6 +1881,105 @@ void inner_rx_64qam_2layer (NR_DL_FRAME_PARMS *frame_parms,
nr_ulsch_qam64_qam64((c16_t *)rxFext_comp[1], (c16_t *)rxFext_comp[0], ul_ch_mag1, ul_ch_mag0, llr_1, rho1, length); nr_ulsch_qam64_qam64((c16_t *)rxFext_comp[1], (c16_t *)rxFext_comp[0], ul_ch_mag1, ul_ch_mag0, llr_1, rho1, length);
} }
void inner_rx_compensation_mag_mrc(int *rxF,
int *ul_ch,
int *rxF_comp,
int *rxF_ch_mag,
int *rxF_ch_magb,
int *rxF_ch_magc,
uint8_t m_order,
int aarx,
int length,
int output_shift) {
register simde__m128i xmmp0, xmmp1, xmmp2, xmmp3, xmmp4;
register simde__m128i complex_shuffle128 = simde_mm_set_epi8(13, 12, 15, 14, 9, 8, 11, 10, 5, 4, 7, 6, 1, 0, 3, 2);
register simde__m128i conj128 = simde_mm_set_epi16(1, -1, 1, -1, 1, -1, 1, -1);
simde__m128i *rxF128 = (simde__m128i*)rxF;
simde__m128i *rxF128_comp = (simde__m128i*)rxF_comp;
simde__m128i *ulch128 = (simde__m128i*)ul_ch;
simde__m128i *ch128_mags[3];
ch128_mags[0] = (simde__m128i*) rxF_ch_mag;
ch128_mags[1] = (simde__m128i*) rxF_ch_magb;
ch128_mags[2] = (simde__m128i*) rxF_ch_magc;
// Prepare constant magnitudes for m_order > 2
simde__m128i QAM_amp128;
simde__m128i QAM_amp128b;
simde__m128i QAM_amp128c;
if(m_order == 4) {
QAM_amp128 = simde_mm_set1_epi16(QAM16_n1);
}
else if(m_order == 6) {
QAM_amp128 = simde_mm_set1_epi16(QAM64_n1); // 4/sqrt(42)
QAM_amp128b = simde_mm_set1_epi16(QAM64_n2); // 2/sqrt(42)
}
else if(m_order == 8){
QAM_amp128 = simde_mm_set1_epi16(QAM256_n1);
QAM_amp128b = simde_mm_set1_epi16(QAM256_n2);
QAM_amp128c = simde_mm_set1_epi16(QAM256_n3);
}
// Loop by every 4 REs
const int num_iter = (length >> 2) + ((length & 3) > 0 ? 1 : 0);
for (int i = 0; i < num_iter; i++) {
// Channel compensation
xmmp0 = simde_mm_madd_epi16(ulch128[i], rxF128[i]);
// xmmp0 contains real part of 4 consecutive outputs (32-bit) of conj(H_m[i])*R_m[i]
xmmp1 = simde_mm_shuffle_epi8(ulch128[i], complex_shuffle128);
xmmp1 = simde_mm_sign_epi16(xmmp1, conj128);
xmmp1 = simde_mm_madd_epi16(xmmp1, rxF128[i]);
// xmmp1 contains imag part of 4 consecutive outputs (32-bit) of conj(H_m[i])*R_m[i]
xmmp0 = simde_mm_srai_epi32(xmmp0, output_shift);
xmmp1 = simde_mm_srai_epi32(xmmp1, output_shift);
xmmp2 = simde_mm_unpacklo_epi32(xmmp0, xmmp1);
xmmp3 = simde_mm_unpackhi_epi32(xmmp0, xmmp1);
xmmp4 = simde_mm_packs_epi32(xmmp2, xmmp3); // xmmp4 is the compensated data
// Compute channel amplitude for LLR computation later
// ch_mag: xmmp1, ch_magb: xmmp2, ch_magc: xmmp3
if(m_order > 2){
xmmp0 = simde_mm_madd_epi16(ulch128[i], ulch128[i]); // |h|^2
xmmp0 = simde_mm_srai_epi32(xmmp0, output_shift);
xmmp0 = simde_mm_packs_epi32(xmmp0, xmmp0);
xmmp0 = simde_mm_unpacklo_epi16(xmmp0, xmmp0);
// Using switch case seems no faster
if(m_order == 4) {
xmmp1 = simde_mm_mulhrs_epi16(xmmp0, QAM_amp128);
}
else if(m_order == 6) {
xmmp1 = simde_mm_mulhrs_epi16(xmmp0, QAM_amp128);
xmmp2 = simde_mm_mulhrs_epi16(xmmp0, QAM_amp128b);
}
else if(m_order == 8){
xmmp1 = simde_mm_mulhrs_epi16(xmmp0, QAM_amp128);
xmmp2 = simde_mm_mulhrs_epi16(xmmp0, QAM_amp128b);
xmmp3 = simde_mm_mulhrs_epi16(xmmp0, QAM_amp128c);
}
}
// MRC (Maximal Ratio Combining, combining compensated signals and channel magnitude)
if (aarx == 0) {
rxF128_comp[i] = xmmp4;
if(m_order >= 4)
ch128_mags[0][i] = xmmp1;
if(m_order >= 6)
ch128_mags[1][i] = xmmp2;
if(m_order == 8)
ch128_mags[2][i] = xmmp3;
}
else {
rxF128_comp[i] = simde_mm_add_epi16(rxF128_comp[i], xmmp4);
if(m_order >= 4)
ch128_mags[0][i] = simde_mm_add_epi16(ch128_mags[0][i], xmmp1);
if(m_order >= 6)
ch128_mags[1][i] = simde_mm_add_epi16(ch128_mags[1][i], xmmp2);
if(m_order == 8)
ch128_mags[2][i] = simde_mm_add_epi16(ch128_mags[2][i], xmmp3);
}
}
}
void inner_rx_qpsk (int *rxF, void inner_rx_qpsk (int *rxF,
int *ul_ch, int *ul_ch,
int16_t *llr, int16_t *llr,
...@@ -2617,7 +2716,7 @@ void nr_pusch_symbol_processing_noprecoding(void *arg) ...@@ -2617,7 +2716,7 @@ void nr_pusch_symbol_processing_noprecoding(void *arg)
int ulsch_id = rdata->ulsch_id; int ulsch_id = rdata->ulsch_id;
int slot = rdata->slot; int slot = rdata->slot;
NR_gNB_PUSCH *pusch_vars = &gNB->pusch_vars[ulsch_id]; NR_gNB_PUSCH *pusch_vars = &gNB->pusch_vars[ulsch_id];
// int16_t *s = rdata->s; int16_t *s = rdata->s;
for (int symbol = rdata->startSymbol; symbol < rdata->startSymbol+rdata->numSymbols; symbol++) for (int symbol = rdata->startSymbol; symbol < rdata->startSymbol+rdata->numSymbols; symbol++)
{ {
int dmrs_symbol_flag = (rel15_ul->ul_dmrs_symb_pos >> symbol) & 0x01; int dmrs_symbol_flag = (rel15_ul->ul_dmrs_symb_pos >> symbol) & 0x01;
...@@ -2639,18 +2738,21 @@ void nr_pusch_symbol_processing_noprecoding(void *arg) ...@@ -2639,18 +2738,21 @@ void nr_pusch_symbol_processing_noprecoding(void *arg)
if (rel15_ul->nrOfLayers == 1) if (rel15_ul->nrOfLayers == 1)
{ {
int16_t *llr = &rdata->llr[0][pusch_vars->llr_offset[symbol]]; int16_t *llr = &rdata->llr[0][pusch_vars->llr_offset[symbol]];
void (*inner_rx)(int *,int *,int16_t *,int,int,int); // void (*inner_rx)(int *,int *,int16_t *,int,int,int);
if (rel15_ul->qam_mod_order == 2) inner_rx = inner_rx_qpsk; // if (rel15_ul->qam_mod_order == 2) inner_rx = inner_rx_qpsk;
else if (rel15_ul->qam_mod_order == 4) inner_rx = inner_rx_16qam; // else if (rel15_ul->qam_mod_order == 4) inner_rx = inner_rx_16qam;
else if (rel15_ul->qam_mod_order == 6) inner_rx = inner_rx_64qam; // else if (rel15_ul->qam_mod_order == 6) inner_rx = inner_rx_64qam;
else if (rel15_ul->qam_mod_order == 8) inner_rx = inner_rx_256qam; // else if (rel15_ul->qam_mod_order == 8) inner_rx = inner_rx_256qam;
else AssertFatal(1==0,"rel15_ul->qam_mod_order %d, pusch_pdu->dmrs_config_type %d\n", // else AssertFatal(1==0,"rel15_ul->qam_mod_order %d, pusch_pdu->dmrs_config_type %d\n",
rel15_ul->qam_mod_order,rel15_ul->dmrs_config_type); // rel15_ul->qam_mod_order,rel15_ul->dmrs_config_type);
int soffset = (slot&3)*frame_parms->symbols_per_slot*frame_parms->ofdm_symbol_size; int soffset = (slot&3)*frame_parms->symbols_per_slot*frame_parms->ofdm_symbol_size;
int32_t rxFext[nb_re_pusch+8] __attribute__((aligned(32))); int32_t rxFext[nb_re_pusch+8] __attribute__((aligned(32)));
int32_t rxF_comp[nb_re_pusch+8] __attribute__((aligned(32)));
int32_t rxF_ch_mags[3][nb_re_pusch+8] __attribute__((aligned(32)));; // rxF_ch_mags[0..2] for ul_ch_mag,ul_ch_magb,ul_ch_magc
int32_t chFext[nb_re_pusch+8] __attribute__((aligned(32))); int32_t chFext[nb_re_pusch+8] __attribute__((aligned(32)));
// int16_t llr16[(nb_re_pusch*rel15_ul->qam_mod_order)+16] __attribute__((aligned(32))); int16_t llr_temp[(nb_re_pusch*rel15_ul->qam_mod_order)+16] __attribute__((aligned(32)));
int16_t *llr_ptr = llr_temp;
for (int aa=0;aa<frame_parms->nb_antennas_rx;aa++) { for (int aa=0;aa<frame_parms->nb_antennas_rx;aa++) {
nr_ulsch_extract_rbs0(gNB->common_vars.rxdataF[aa], nr_ulsch_extract_rbs0(gNB->common_vars.rxdataF[aa],
gNB->pusch_vars[ulsch_id].ul_ch_estimates[aa], gNB->pusch_vars[ulsch_id].ul_ch_estimates[aa],
...@@ -2662,16 +2764,49 @@ void nr_pusch_symbol_processing_noprecoding(void *arg) ...@@ -2662,16 +2764,49 @@ void nr_pusch_symbol_processing_noprecoding(void *arg)
dmrs_symbol_flag, dmrs_symbol_flag,
rel15_ul, rel15_ul,
frame_parms); frame_parms);
// demodulation // demodulation
inner_rx(rxFext, chFext, llr, aa, nb_re_pusch, gNB->pusch_vars[ulsch_id].log2_maxh); // inner_rx(rxFext, chFext, llr_temp, aa, nb_re_pusch, gNB->pusch_vars[ulsch_id].log2_maxh);
// Channel compensation, get channel magnitude and MRC
inner_rx_compensation_mag_mrc(rxFext,
chFext,
rxF_comp,
rxF_ch_mags[0],
rxF_ch_mags[1],
rxF_ch_mags[2],
rel15_ul->qam_mod_order,
aa,
nb_re_pusch,
gNB->pusch_vars[ulsch_id].log2_maxh
);
} }
// Perform IDFT if transform precoding is enabled
// TODO: Frequency equalizer
if(rel15_ul->transform_precoding == transformPrecoder_enabled){
nr_idft(rxF_comp, nb_re_pusch);
}
// LLR computation
if(rel15_ul->qam_mod_order == 2)
llr_ptr = (int16_t*) rxF_comp; // dont call llr function if it's QPSK
else
nr_ulsch_compute_llr(rxF_comp,
rxF_ch_mags[0],
rxF_ch_mags[1],
rxF_ch_mags[2],
llr_temp,
0, // nb_re=0 cuz no offset in ch_mags[]
nb_re_pusch,
0, // symbol=0 cuz no offset in ch_mags[]
rel15_ul->qam_mod_order);
// unscrambling // unscrambling
// simde__m64 *llr64 = (simde__m64 *) llr; simde__m64 *llr64 = (simde__m64 *) llr;
// for (int i=0;i<(nb_re_pusch*rel15_ul->qam_mod_order)>>2;i++) { for (int i=0;i<(nb_re_pusch*rel15_ul->qam_mod_order)>>2;i++) {
// llr64[i] = simde_mm_mullo_pi16(((simde__m64 *)llr16)[i],((simde__m64 *)s)[i]); llr64[i] = simde_mm_mullo_pi16(((simde__m64 *)llr_ptr)[i],((simde__m64 *)s)[i]);
// } }
// s+=(nb_re_pusch*rel15_ul->qam_mod_order); s+=(nb_re_pusch*rel15_ul->qam_mod_order);
// llr+=(nb_re_pusch*rel15_ul->qam_mod_order); llr+=(nb_re_pusch*rel15_ul->qam_mod_order);
} }
else else
{ {
......
openair1/SCHED_NR/phy_procedures_nr_gNB.c
View file @ c35025f5
...@@ -441,20 +441,20 @@ void nr_ulsch_procedures(PHY_VARS_gNB *gNB, int frame_rx, int slot_rx, int ULSCH ...@@ -441,20 +441,20 @@ void nr_ulsch_procedures(PHY_VARS_gNB *gNB, int frame_rx, int slot_rx, int ULSCH
pusch_pdu->nrOfLayers, pusch_pdu->nrOfLayers,
G); G);
// if (gNB->use_pusch_tp == 0) { nr_ulsch_layer_demapping(gNB->pusch_vars[ULSCH_id].llr,
nr_ulsch_layer_demapping(gNB->pusch_vars[ULSCH_id].llr, pusch_pdu->nrOfLayers,
pusch_pdu->nrOfLayers, pusch_pdu->qam_mod_order,
pusch_pdu->qam_mod_order, G,
G, gNB->pusch_vars[ULSCH_id].llr_layers);
gNB->pusch_vars[ULSCH_id].llr_layers);
if (gNB->use_pusch_tp == 0 || (gNB->use_pusch_tp == 1 && pusch_pdu->nrOfLayers > 1)) {
//---------------------------------------------------------- //----------------------------------------------------------
//------------------- ULSCH unscrambling ------------------- //------------------- ULSCH unscrambling -------------------
//---------------------------------------------------------- //----------------------------------------------------------
start_meas(&gNB->ulsch_unscrambling_stats); start_meas(&gNB->ulsch_unscrambling_stats);
nr_ulsch_unscrambling(gNB->pusch_vars[ULSCH_id].llr, G, pusch_pdu->data_scrambling_id, pusch_pdu->rnti); nr_ulsch_unscrambling(gNB->pusch_vars[ULSCH_id].llr, G, pusch_pdu->data_scrambling_id, pusch_pdu->rnti);
stop_meas(&gNB->ulsch_unscrambling_stats); stop_meas(&gNB->ulsch_unscrambling_stats);
// } }
//---------------------------------------------------------- //----------------------------------------------------------
//--------------------- ULSCH decoding --------------------- //--------------------- ULSCH decoding ---------------------
//---------------------------------------------------------- //----------------------------------------------------------
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment