dci.c 117 KB
Newer Older
1
/*******************************************************************************
2
    OpenAirInterface
ghaddab's avatar
ghaddab committed
3
    Copyright(c) 1999 - 2014 Eurecom
4

ghaddab's avatar
ghaddab committed
5 6 7 8
    OpenAirInterface is free software: you can redistribute it and/or modify
    it under the terms of the GNU General Public License as published by
    the Free Software Foundation, either version 3 of the License, or
    (at your option) any later version.
9 10


ghaddab's avatar
ghaddab committed
11 12 13 14
    OpenAirInterface is distributed in the hope that it will be useful,
    but WITHOUT ANY WARRANTY; without even the implied warranty of
    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
    GNU General Public License for more details.
15

ghaddab's avatar
ghaddab committed
16
    You should have received a copy of the GNU General Public License
17 18
    along with OpenAirInterface.The full GNU General Public License is
   included in this distribution in the file called "COPYING". If not,
ghaddab's avatar
ghaddab committed
19
   see <http://www.gnu.org/licenses/>.
20 21

  Contact Information
ghaddab's avatar
ghaddab committed
22 23 24
  OpenAirInterface Admin: openair_admin@eurecom.fr
  OpenAirInterface Tech : openair_tech@eurecom.fr
  OpenAirInterface Dev  : openair4g-devel@eurecom.fr
25

ghaddab's avatar
ghaddab committed
26
  Address      : Eurecom, Campus SophiaTech, 450 Route des Chappes, CS 50193 - 06904 Biot Sophia Antipolis cedex, FRANCE
27

ghaddab's avatar
ghaddab committed
28
 *******************************************************************************/
29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47

/*! \file PHY/LTE_TRANSPORT/dci.c
* \brief Implements PDCCH physical channel TX/RX procedures (36.211) and DCI encoding/decoding (36.212/36.213). Current LTE compliance V8.6 2009-03.
* \author R. Knopp
* \date 2011
* \version 0.1
* \company Eurecom
* \email: knopp@eurecom.fr
* \note
* \warning
*/
#ifdef USER_MODE
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#endif
#include "PHY/defs.h"
#include "PHY/extern.h"
#include "SCHED/defs.h"
48
#include "SIMULATION/TOOLS/defs.h" // for taus 
49
#include "PHY/sse_intrin.h"
50

51
#include "assertions.h"
52 53 54 55

//#define DEBUG_DCI_ENCODING 1
//#define DEBUG_DCI_DECODING 1
//#define DEBUG_PHY
56

57 58
//#undef ALL_AGGREGATION

59 60
//extern uint16_t phich_reg[MAX_NUM_PHICH_GROUPS][3];
//extern uint16_t pcfich_reg[4];
61

62 63
uint32_t check_phich_reg(LTE_DL_FRAME_PARMS *frame_parms,uint32_t kprime,uint8_t lprime,uint8_t mi)
{
64

65 66 67 68
  uint16_t i;
  uint16_t Ngroup_PHICH = (frame_parms->phich_config_common.phich_resource*frame_parms->N_RB_DL)/48;
  uint16_t mprime;
  uint16_t *pcfich_reg = frame_parms->pcfich_reg;
69 70 71 72 73 74 75 76 77 78 79 80 81 82

  if ((lprime>0) && (frame_parms->Ncp==0) )
    return(0);

  //  printf("check_phich_reg : mi %d\n",mi);

  // compute REG based on symbol
  if ((lprime == 0)||
      ((lprime==1)&&(frame_parms->nb_antennas_tx_eNB == 4)))
    mprime = kprime/6;
  else
    mprime = kprime>>2;

  // check if PCFICH uses mprime
83
  if ((lprime==0) &&
84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99
      ((mprime == pcfich_reg[0]) ||
       (mprime == pcfich_reg[1]) ||
       (mprime == pcfich_reg[2]) ||
       (mprime == pcfich_reg[3]))) {
#ifdef DEBUG_DCI_ENCODING
    msg("[PHY] REG %d allocated to PCFICH\n",mprime);
#endif
    return(1);
  }

  // handle Special subframe case for TDD !!!

  //  printf("Checking phich_reg %d\n",mprime);
  if (mi > 0) {
    if (((frame_parms->phich_config_common.phich_resource*frame_parms->N_RB_DL)%48) > 0)
      Ngroup_PHICH++;
100

101 102 103
    if (frame_parms->Ncp == 1) {
      Ngroup_PHICH<<=1;
    }
104 105 106 107 108 109 110



    for (i=0; i<Ngroup_PHICH; i++) {
      if ((mprime == frame_parms->phich_reg[i][0]) ||
          (mprime == frame_parms->phich_reg[i][1]) ||
          (mprime == frame_parms->phich_reg[i][2]))  {
111
#ifdef DEBUG_DCI_ENCODING
112
        msg("[PHY] REG %d (lprime %d) allocated to PHICH\n",mprime,lprime);
113
#endif
114
        return(1);
115 116 117
      }
    }
  }
118

119 120 121
  return(0);
}

122 123
uint16_t extract_crc(uint8_t *dci,uint8_t dci_len)
{
124

125 126
  uint16_t crc16;
  //  uint8_t i;
127 128

  /*
129 130
  uint8_t crc;
  crc = ((uint16_t *)dci)[DCI_LENGTH>>4];
131 132 133
  printf("crc1: %x, shift %d (DCI_LENGTH %d)\n",crc,DCI_LENGTH&0xf,DCI_LENGTH);
  crc = (crc>>(DCI_LENGTH&0xf));
  // clear crc bits
134 135 136
  ((uint16_t *)dci)[DCI_LENGTH>>4] &= (0xffff>>(16-(DCI_LENGTH&0xf)));
  printf("crc2: %x, dci0 %x\n",crc,((int16_t *)dci)[DCI_LENGTH>>4]);
  crc |= (((uint16_t *)dci)[1+(DCI_LENGTH>>4)])<<(16-(DCI_LENGTH&0xf));
137
  // clear crc bits
138
  (((uint16_t *)dci)[1+(DCI_LENGTH>>4)]) = 0;
139 140
  printf("extract_crc: crc %x\n",crc);
  */
141
#ifdef DEBUG_DCI_DECODING
142
  msg("dci_crc (%x,%x,%x), dci_len&0x7=%d\n",dci[dci_len>>3],dci[1+(dci_len>>3)],dci[2+(dci_len>>3)],
143
      dci_len&0x7);
144
#endif
145

146
  if ((dci_len&0x7) > 0) {
147 148
    ((uint8_t *)&crc16)[0] = dci[1+(dci_len>>3)]<<(dci_len&0x7) | dci[2+(dci_len>>3)]>>(8-(dci_len&0x7));
    ((uint8_t *)&crc16)[1] = dci[(dci_len>>3)]<<(dci_len&0x7) | dci[1+(dci_len>>3)]>>(8-(dci_len&0x7));
149
  } else {
150 151
    ((uint8_t *)&crc16)[0] = dci[1+(dci_len>>3)];
    ((uint8_t *)&crc16)[1] = dci[(dci_len>>3)];
152 153
  }

154
#ifdef DEBUG_DCI_DECODING
155 156 157 158 159 160
  msg("dci_crc =>%x\n",crc16);
#endif

  //  dci[(dci_len>>3)]&=(0xffff<<(dci_len&0xf));
  //  dci[(dci_len>>3)+1] = 0;
  //  dci[(dci_len>>3)+2] = 0;
161
  return((uint16_t)crc16);
162 163 164 165 166

}



167 168
static uint8_t d[3*(MAX_DCI_SIZE_BITS + 16) + 96];
static uint8_t w[3*3*(MAX_DCI_SIZE_BITS+16)];
169

170
void dci_encoding(uint8_t *a,
171 172 173 174 175
                  uint8_t A,
                  uint16_t E,
                  uint8_t *e,
                  uint16_t rnti)
{
176 177


178 179
  uint8_t D = (A + 16);
  uint32_t RCC;
180 181

#ifdef DEBUG_DCI_ENCODING
182
  int32_t i;
183
#endif
184
  // encode dci
185 186 187 188 189 190 191 192 193 194

#ifdef DEBUG_DCI_ENCODING
  msg("Doing DCI encoding for %d bits, e %p, rnti %x\n",A,e,rnti);
#endif

  memset((void *)d,LTE_NULL,96);

  ccodelte_encode(A,2,a,d+96,rnti);

#ifdef DEBUG_DCI_ENCODING
195 196

  for (i=0; i<16+A; i++)
197
    msg("%d : (%d,%d,%d)\n",i,*(d+96+(3*i)),*(d+97+(3*i)),*(d+98+(3*i)));
198

199
#endif
200

201 202 203 204 205 206 207 208 209 210 211 212 213 214
#ifdef DEBUG_DCI_ENCODING
  msg("Doing DCI interleaving for %d coded bits, e %p\n",D*3,e);
#endif
  RCC = sub_block_interleaving_cc(D,d+96,w);

#ifdef DEBUG_DCI_ENCODING
  msg("Doing DCI rate matching for %d channel bits, RCC %d, e %p\n",E,RCC,e);
#endif
  lte_rate_matching_cc(RCC,E,w,e);


}


215
uint8_t *generate_dci0(uint8_t *dci,
216 217 218 219 220 221
                       uint8_t *e,
                       uint8_t DCI_LENGTH,
                       uint8_t aggregation_level,
                       uint16_t rnti)
{

222 223
  uint16_t coded_bits;
  uint8_t dci_flip[8];
224 225 226 227 228 229 230 231

  if (aggregation_level>3) {
    msg("dci.c: generate_dci FATAL, illegal aggregation_level %d\n",aggregation_level);
    return NULL;
  }

  coded_bits = 72 * (1<<aggregation_level);

232 233 234
  /*

  #ifdef DEBUG_DCI_ENCODING
235 236
  for (i=0;i<1+((DCI_LENGTH+16)/8);i++)
    msg("i %d : %x\n",i,dci[i]);
237
  #endif
238
  */
239
  if (DCI_LENGTH<=32) {
240 241 242
    dci_flip[0] = dci[3];
    dci_flip[1] = dci[2];
    dci_flip[2] = dci[1];
243 244
    dci_flip[3] = dci[0];
  } else {
245 246 247 248 249 250 251 252
    dci_flip[0] = dci[7];
    dci_flip[1] = dci[6];
    dci_flip[2] = dci[5];
    dci_flip[3] = dci[4];
    dci_flip[4] = dci[3];
    dci_flip[5] = dci[2];
    dci_flip[6] = dci[1];
    dci_flip[7] = dci[0];
253
#ifdef DEBUG_DCI_ENCODING
254 255 256
    msg("DCI => %x,%x,%x,%x,%x,%x,%x,%x\n",
        dci_flip[0],dci_flip[1],dci_flip[2],dci_flip[3],
        dci_flip[4],dci_flip[5],dci_flip[6],dci_flip[7]);
257
#endif
258
  }
259

260 261 262 263 264
  dci_encoding(dci_flip,DCI_LENGTH,coded_bits,e,rnti);

  return(e+coded_bits);
}

265
uint32_t Y;
266 267 268 269 270 271 272 273

#define CCEBITS 72
#define CCEPERSYMBOL 33  // This is for 1200 RE
#define CCEPERSYMBOL0 22  // This is for 1200 RE
#define DCI_BITS_MAX ((2*CCEPERSYMBOL+CCEPERSYMBOL0)*CCEBITS)
#define Msymb (DCI_BITS_MAX/2)
//#define Mquad (Msymb/4)

274
static uint32_t bitrev_cc_dci[32] = {1,17,9,25,5,21,13,29,3,19,11,27,7,23,15,31,0,16,8,24,4,20,12,28,2,18,10,26,6,22,14,30};
275 276
static mod_sym_t wtemp[2][Msymb];

277 278
void pdcch_interleaving(LTE_DL_FRAME_PARMS *frame_parms,mod_sym_t **z, mod_sym_t **wbar,uint8_t n_symbols_pdcch,uint8_t mi)
{
279 280

  mod_sym_t *wptr,*wptr2,*zptr;
281 282 283 284
  uint32_t Mquad = get_nquad(n_symbols_pdcch,frame_parms,mi);
  uint32_t RCC = (Mquad>>5), ND;
  uint32_t row,col,Kpi,index;
  int32_t i,k,a;
285
#ifdef RM_DEBUG
286
  int32_t nulled=0;
287
#endif
288

289 290 291
  //  msg("[PHY] PDCCH Interleaving Mquad %d (Nsymb %d)\n",Mquad,n_symbols_pdcch);
  if ((Mquad&0x1f) > 0)
    RCC++;
292

293 294 295 296
  Kpi = (RCC<<5);
  ND = Kpi - Mquad;

  k=0;
297 298

  for (col=0; col<32; col++) {
299 300
    index = bitrev_cc_dci[col];

301
    for (row=0; row<RCC; row++) {
302 303
      //msg("col %d, index %d, row %d\n",col,index,row);
      if (index>=ND) {
304 305 306 307 308
        for (a=0; a<frame_parms->nb_antennas_tx_eNB; a++) {
          //msg("a %d k %d\n",a,k);

          wptr = &wtemp[a][k<<2];
          zptr = &z[a][(index-ND)<<2];
309

310
          //msg("wptr=%p, zptr=%p\n",wptr,zptr);
311

312 313 314 315 316
          wptr[0] = zptr[0];
          wptr[1] = zptr[1];
          wptr[2] = zptr[2];
          wptr[3] = zptr[3];
        }
317

318
        k++;
319
      }
320

321 322 323 324 325
      index+=32;
    }
  }

  // permutation
326 327 328
  for (i=0; i<Mquad; i++) {

    for (a=0; a<frame_parms->nb_antennas_tx_eNB; a++) {
329 330 331 332 333 334 335 336 337 338 339 340 341

      //wptr  = &wtemp[a][i<<2];
      //wptr2 = &wbar[a][((i+frame_parms->Nid_cell)%Mquad)<<2];
      wptr = &wtemp[a][((i+frame_parms->Nid_cell)%Mquad)<<2];
      wptr2 = &wbar[a][i<<2];
      wptr2[0] = wptr[0];
      wptr2[1] = wptr[1];
      wptr2[2] = wptr[2];
      wptr2[3] = wptr[3];
    }
  }
}

342 343
void pdcch_demapping(uint16_t *llr,uint16_t *wbar,LTE_DL_FRAME_PARMS *frame_parms,uint8_t num_pdcch_symbols,uint8_t mi)
{
344

345 346 347
  uint32_t i, lprime;
  uint16_t kprime,kprime_mod12,mprime,symbol_offset,tti_offset,tti_offset0;
  int16_t re_offset,re_offset0;
348 349 350 351 352 353 354 355 356

  // This is the REG allocation algorithm from 36-211, second part of Section 6.8.5

  int Msymb2;

  switch (frame_parms->N_RB_DL) {
  case 100:
    Msymb2 = Msymb;
    break;
357

358 359 360
  case 75:
    Msymb2 = 3*Msymb/4;
    break;
361

362 363 364
  case 50:
    Msymb2 = Msymb>>1;
    break;
365

366 367 368
  case 25:
    Msymb2 = Msymb>>2;
    break;
369

370 371 372
  case 15:
    Msymb2 = Msymb*15/100;
    break;
373

374 375 376
  case 6:
    Msymb2 = Msymb*6/100;
    break;
377

378 379 380 381
  default:
    Msymb2 = Msymb>>2;
    break;
  }
382

383 384 385 386 387
  mprime=0;


  re_offset = 0;
  re_offset0 = 0; // counter for symbol with pilots (extracted outside!)
388 389 390

  for (kprime=0; kprime<frame_parms->N_RB_DL*12; kprime++) {
    for (lprime=0; lprime<num_pdcch_symbols; lprime++) {
391

392
      symbol_offset = (uint32_t)frame_parms->N_RB_DL*12*lprime;
393

394 395
      tti_offset = symbol_offset + re_offset;
      tti_offset0 = symbol_offset + re_offset0;
396

397 398
      // if REG is allocated to PHICH, skip it
      if (check_phich_reg(frame_parms,kprime,lprime,mi) == 1) {
399 400 401 402 403 404 405 406 407 408 409 410 411 412
        //msg("dci_demapping : skipping REG %d\n",(lprime==0)?kprime/6 : kprime>>2);
        if ((lprime == 0)&&((kprime%6)==0))
          re_offset0+=4;
      } else { // not allocated to PHICH/PCFICH
        //  msg("dci_demapping: REG %d\n",(lprime==0)?kprime/6 : kprime>>2);
        if (lprime == 0) {
          // first symbol, or second symbol+4 TX antennas skip pilots
          kprime_mod12 = kprime%12;

          if ((kprime_mod12 == 0) || (kprime_mod12 == 6)) {
            // kprime represents REG

            for (i=0; i<4; i++) {
              wbar[mprime] = llr[tti_offset0+i];
413
#ifdef DEBUG_DCI_DECODING
414
              msg("[PHY] PDCCH demapping mprime %d => %d (symbol %d re %d) -> (%d,%d)\n",mprime,tti_offset0,symbol_offset,re_offset0,*(char*)&wbar[mprime],*(1+(char*)&wbar[mprime]));
415
#endif
416 417 418 419 420 421 422 423 424 425 426 427 428
              mprime++;
              re_offset0++;
            }
          }
        } else if ((lprime==1)&&(frame_parms->nb_antennas_tx_eNB == 4)) {
          // LATER!!!!
        } else { // no pilots in this symbol
          kprime_mod12 = kprime%12;

          if ((kprime_mod12 == 0) || (kprime_mod12 == 4) || (kprime_mod12 == 8)) {
            // kprime represents REG
            for (i=0; i<4; i++) {
              wbar[mprime] = llr[tti_offset+i];
429
#ifdef DEBUG_DCI_DECODING
430
              msg("[PHY] PDCCH demapping mprime %d => %d (symbol %d re %d) -> (%d,%d)\n",mprime,tti_offset,symbol_offset,re_offset+i,*(char*)&wbar[mprime],*(1+(char*)&wbar[mprime]));
431
#endif
432 433 434 435
              mprime++;
            }
          }  // is representative
        } // no pilots case
436 437 438 439
      } // not allocated to PHICH/PCFICH

      // Stop when all REGs are copied in
      if (mprime>=Msymb2)
440
        break;
441
    } //lprime loop
442

443 444 445 446 447
    re_offset++;

  } // kprime loop
}

448
static uint16_t wtemp_rx[Msymb];
449 450
void pdcch_deinterleaving(LTE_DL_FRAME_PARMS *frame_parms,uint16_t *z, uint16_t *wbar,uint8_t number_pdcch_symbols,uint8_t mi)
{
451

452
  uint16_t *wptr,*zptr,*wptr2;
453

454 455 456 457
  uint16_t Mquad=get_nquad(number_pdcch_symbols,frame_parms,mi);
  uint32_t RCC = (Mquad>>5), ND;
  uint32_t row,col,Kpi,index;
  int32_t i,k;
458 459 460 461 462 463 464 465


  //  printf("Mquad %d, RCC %d\n",Mquad,RCC);

  if (!z) {
    msg("dci.c: pdcch_deinterleaving: FATAL z is Null\n");
    return;
  }
466

467
  // undo permutation
468
  for (i=0; i<Mquad; i++) {
469 470 471 472 473 474 475 476 477 478 479
    //wptr = &wtemp_rx[i<<2];
    //wptr2 = &wbar[((i+frame_parms->Nid_cell)%Mquad)<<2];
    wptr = &wtemp_rx[((i+frame_parms->Nid_cell)%Mquad)<<2];
    wptr2 = &wbar[i<<2];

    wptr[0] = wptr2[0];
    wptr[1] = wptr2[1];
    wptr[2] = wptr2[2];
    wptr[3] = wptr2[3];
    /*
    msg("pdcch_deinterleaving (%p,%p): quad %d -> (%d,%d %d,%d %d,%d %d,%d)\n",wptr,wptr2,i,
480 481 482 483 484 485 486 487 488
    ((char*)wptr2)[0],
    ((char*)wptr2)[1],
    ((char*)wptr2)[2],
    ((char*)wptr2)[3],
    ((char*)wptr2)[4],
    ((char*)wptr2)[5],
    ((char*)wptr2)[6],
    ((char*)wptr2)[7]);
    */
489 490 491 492 493

  }

  if ((Mquad&0x1f) > 0)
    RCC++;
494

495 496 497 498
  Kpi = (RCC<<5);
  ND = Kpi - Mquad;

  k=0;
499 500

  for (col=0; col<32; col++) {
501 502
    index = bitrev_cc_dci[col];

503
    for (row=0; row<RCC; row++) {
504 505 506 507 508
      //      printf("row %d, index %d, Nd %d\n",row,index,ND);
      if (index>=ND) {



509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528
        wptr = &wtemp_rx[k<<2];
        zptr = &z[(index-ND)<<2];

        zptr[0] = wptr[0];
        zptr[1] = wptr[1];
        zptr[2] = wptr[2];
        zptr[3] = wptr[3];

        /*
        printf("deinterleaving ; k %d, index-Nd %d  => (%d,%d,%d,%d,%d,%d,%d,%d)\n",k,(index-ND),
               ((int8_t *)wptr)[0],
               ((int8_t *)wptr)[1],
               ((int8_t *)wptr)[2],
               ((int8_t *)wptr)[3],
               ((int8_t *)wptr)[4],
               ((int8_t *)wptr)[5],
               ((int8_t *)wptr)[6],
               ((int8_t *)wptr)[7]);
        */
        k++;
529
      }
530

531
      index+=32;
532

533 534 535
    }
  }

536
  for (i=0; i<Mquad; i++) {
537 538 539
    zptr = &z[i<<2];
    /*
    printf("deinterleaving ; quad %d  => (%d,%d,%d,%d,%d,%d,%d,%d)\n",i,
540 541 542 543 544 545 546 547
     ((int8_t *)zptr)[0],
     ((int8_t *)zptr)[1],
     ((int8_t *)zptr)[2],
     ((int8_t *)zptr)[3],
     ((int8_t *)zptr)[4],
     ((int8_t *)zptr)[5],
     ((int8_t *)zptr)[6],
     ((int8_t *)zptr)[7]);
548 549
    */
  }
550

551 552 553
}


554
int32_t pdcch_qpsk_qpsk_llr(LTE_DL_FRAME_PARMS *frame_parms,
555 556 557 558 559 560 561
                            int32_t **rxdataF_comp,
                            int32_t **rxdataF_comp_i,
                            int32_t **rho_i,
                            int16_t *pdcch_llr16,
                            int16_t *pdcch_llr8in,
                            uint8_t symbol)
{
562 563 564 565 566

  __m128i *rxF=(__m128i*)&rxdataF_comp[0][(symbol*frame_parms->N_RB_DL*12)];
  __m128i *rxF_i=(__m128i*)&rxdataF_comp_i[0][(symbol*frame_parms->N_RB_DL*12)];
  __m128i *rho=(__m128i*)&rho_i[0][(symbol*frame_parms->N_RB_DL*12)];
  __m128i *llr128;
567
  int32_t i;
568
  char *pdcch_llr8;
569
  int16_t *pdcch_llr;
570 571 572 573
  pdcch_llr8 = (char *)&pdcch_llr8in[symbol*frame_parms->N_RB_DL*12];
  pdcch_llr = &pdcch_llr16[symbol*frame_parms->N_RB_DL*12];

  //  printf("dlsch_qpsk_qpsk: symbol %d\n",symbol);
574

575 576 577 578 579 580 581
  llr128 = (__m128i*)pdcch_llr;

  if (!llr128) {
    msg("dlsch_qpsk_qpsk_llr: llr is null, symbol %d\n",symbol);
    return -1;
  }

582
  qpsk_qpsk((int16_t *)rxF,
583 584 585 586
            (int16_t *)rxF_i,
            (int16_t *)llr128,
            (int16_t *)rho,
            frame_parms->N_RB_DL*12);
587 588

  //prepare for Viterbi which accepts 8 bit, but prefers 4 bit, soft input.
589
  for (i=0; i<(frame_parms->N_RB_DL*24); i++) {
590 591 592 593 594 595 596 597 598 599 600 601 602 603 604
    if (*pdcch_llr>7)
      *pdcch_llr8=7;
    else if (*pdcch_llr<-8)
      *pdcch_llr8=-8;
    else
      *pdcch_llr8 = (char)(*pdcch_llr);

    pdcch_llr++;
    pdcch_llr8++;
  }

  return(0);
}


605
int32_t pdcch_llr(LTE_DL_FRAME_PARMS *frame_parms,
606 607 608 609
                  int32_t **rxdataF_comp,
                  char *pdcch_llr,
                  uint8_t symbol)
{
610

611 612
  int16_t *rxF= (int16_t*) &rxdataF_comp[0][(symbol*frame_parms->N_RB_DL*12)];
  int32_t i;
613 614 615
  char *pdcch_llr8;

  pdcch_llr8 = &pdcch_llr[2*symbol*frame_parms->N_RB_DL*12];
616

617 618 619 620
  if (!pdcch_llr8) {
    msg("pdcch_qpsk_llr: llr is null, symbol %d\n",symbol);
    return(-1);
  }
621

622 623
  //    msg("pdcch qpsk llr for symbol %d (pos %d), llr offset %d\n",symbol,(symbol*frame_parms->N_RB_DL*12),pdcch_llr8-pdcch_llr);

624
  for (i=0; i<(frame_parms->N_RB_DL*((symbol==0) ? 16 : 24)); i++) {
625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644

    if (*rxF>31)
      *pdcch_llr8=31;
    else if (*rxF<-32)
      *pdcch_llr8=-32;
    else
      *pdcch_llr8 = (char)(*rxF);

    //    printf("%d %d => %d\n",i,*rxF,*pdcch_llr8);
    rxF++;
    pdcch_llr8++;
  }

  return(0);

}

__m128i avg128P;

//compute average channel_level on each (TX,RX) antenna pair
645
void pdcch_channel_level(int32_t **dl_ch_estimates_ext,
646 647 648 649
                         LTE_DL_FRAME_PARMS *frame_parms,
                         int32_t *avg,
                         uint8_t nb_rb)
{
650

651 652
  int16_t rb;
  uint8_t aatx,aarx;
653 654
  __m128i *dl_ch128;

655 656 657

  for (aatx=0; aatx<frame_parms->nb_antennas_tx_eNB; aatx++)
    for (aarx=0; aarx<frame_parms->nb_antennas_rx; aarx++) {
658
      //clear average level
659
      avg128P = _mm_setzero_si128();
660 661
      dl_ch128=(__m128i *)&dl_ch_estimates_ext[(aatx<<1)+aarx][frame_parms->N_RB_DL*12];

662 663 664 665 666 667 668 669 670 671 672 673 674 675
      for (rb=0; rb<nb_rb; rb++) {

        avg128P = _mm_add_epi32(avg128P,_mm_madd_epi16(dl_ch128[0],dl_ch128[0]));
        avg128P = _mm_add_epi32(avg128P,_mm_madd_epi16(dl_ch128[1],dl_ch128[1]));
        avg128P = _mm_add_epi32(avg128P,_mm_madd_epi16(dl_ch128[2],dl_ch128[2]));

        dl_ch128+=3;
        /*
          if (rb==0) {
          print_shorts("dl_ch128",&dl_ch128[0]);
          print_shorts("dl_ch128",&dl_ch128[1]);
          print_shorts("dl_ch128",&dl_ch128[2]);
          }
        */
676 677
      }

678 679 680 681
      avg[(aatx<<1)+aarx] = (((int32_t*)&avg128P)[0] +
                             ((int32_t*)&avg128P)[1] +
                             ((int32_t*)&avg128P)[2] +
                             ((int32_t*)&avg128P)[3])/(nb_rb*12);
682 683 684

      //            msg("Channel level : %d\n",avg[(aatx<<1)+aarx]);
    }
685

686 687 688 689 690 691 692 693
  _mm_empty();
  _m_empty();

}

__m128i mmtmpPD0,mmtmpPD1,mmtmpPD2,mmtmpPD3;

void pdcch_dual_stream_correlation(LTE_DL_FRAME_PARMS *frame_parms,
694 695 696 697 698 699
                                   uint8_t symbol,
                                   int32_t **dl_ch_estimates_ext,
                                   int32_t **dl_ch_estimates_ext_i,
                                   int32_t **dl_ch_rho_ext,
                                   uint8_t output_shift)
{
700

701
  uint16_t rb;
702
  __m128i *dl_ch128,*dl_ch128i,*dl_ch_rho128;
703
  uint8_t aarx;
704 705 706 707

  //  printf("dlsch_dual_stream_correlation: symbol %d\n",symbol);


708
  for (aarx=0; aarx<frame_parms->nb_antennas_rx; aarx++) {
709 710 711 712 713 714

    dl_ch128          = (__m128i *)&dl_ch_estimates_ext[aarx][symbol*frame_parms->N_RB_DL*12];
    dl_ch128i         = (__m128i *)&dl_ch_estimates_ext_i[aarx][symbol*frame_parms->N_RB_DL*12];
    dl_ch_rho128      = (__m128i *)&dl_ch_rho_ext[aarx][symbol*frame_parms->N_RB_DL*12];


715
    for (rb=0; rb<frame_parms->N_RB_DL; rb++) {
716 717
      // multiply by conjugated channel
      mmtmpPD0 = _mm_madd_epi16(dl_ch128[0],dl_ch128i[0]);
718 719
      //  print_ints("re",&mmtmpPD0);

720 721 722 723
      // mmtmpD0 contains real part of 4 consecutive outputs (32-bit)
      mmtmpPD1 = _mm_shufflelo_epi16(dl_ch128[0],_MM_SHUFFLE(2,3,0,1));
      mmtmpPD1 = _mm_shufflehi_epi16(mmtmpPD1,_MM_SHUFFLE(2,3,0,1));
      mmtmpPD1 = _mm_sign_epi16(mmtmpPD1,*(__m128i*)&conjugate[0]);
724
      //  print_ints("im",&mmtmpPD1);
725 726 727
      mmtmpPD1 = _mm_madd_epi16(mmtmpPD1,dl_ch128i[0]);
      // mmtmpD1 contains imag part of 4 consecutive outputs (32-bit)
      mmtmpPD0 = _mm_srai_epi32(mmtmpPD0,output_shift);
728
      //  print_ints("re(shift)",&mmtmpPD0);
729
      mmtmpPD1 = _mm_srai_epi32(mmtmpPD1,output_shift);
730
      //  print_ints("im(shift)",&mmtmpPD1);
731 732
      mmtmpPD2 = _mm_unpacklo_epi32(mmtmpPD0,mmtmpPD1);
      mmtmpPD3 = _mm_unpackhi_epi32(mmtmpPD0,mmtmpPD1);
733 734
      //        print_ints("c0",&mmtmpPD2);
      //  print_ints("c1",&mmtmpPD3);
735
      dl_ch_rho128[0] = _mm_packs_epi32(mmtmpPD2,mmtmpPD3);
736

737 738 739
      //print_shorts("rx:",dl_ch128_2);
      //print_shorts("ch:",dl_ch128);
      //print_shorts("pack:",rho128);
740

741 742 743 744 745 746 747 748 749 750 751 752
      // multiply by conjugated channel
      mmtmpPD0 = _mm_madd_epi16(dl_ch128[1],dl_ch128i[1]);
      // mmtmpPD0 contains real part of 4 consecutive outputs (32-bit)
      mmtmpPD1 = _mm_shufflelo_epi16(dl_ch128[1],_MM_SHUFFLE(2,3,0,1));
      mmtmpPD1 = _mm_shufflehi_epi16(mmtmpPD1,_MM_SHUFFLE(2,3,0,1));
      mmtmpPD1 = _mm_sign_epi16(mmtmpPD1,*(__m128i*)conjugate);
      mmtmpPD1 = _mm_madd_epi16(mmtmpPD1,dl_ch128i[1]);
      // mmtmpPD1 contains imag part of 4 consecutive outputs (32-bit)
      mmtmpPD0 = _mm_srai_epi32(mmtmpPD0,output_shift);
      mmtmpPD1 = _mm_srai_epi32(mmtmpPD1,output_shift);
      mmtmpPD2 = _mm_unpacklo_epi32(mmtmpPD0,mmtmpPD1);
      mmtmpPD3 = _mm_unpackhi_epi32(mmtmpPD0,mmtmpPD1);
753 754


755 756 757
      dl_ch_rho128[1] =_mm_packs_epi32(mmtmpPD2,mmtmpPD3);
      //print_shorts("rx:",dl_ch128_2+1);
      //print_shorts("ch:",dl_ch128+1);
758
      //print_shorts("pack:",rho128+1);
759 760 761 762 763 764 765 766 767 768 769 770
      // multiply by conjugated channel
      mmtmpPD0 = _mm_madd_epi16(dl_ch128[2],dl_ch128i[2]);
      // mmtmpPD0 contains real part of 4 consecutive outputs (32-bit)
      mmtmpPD1 = _mm_shufflelo_epi16(dl_ch128[2],_MM_SHUFFLE(2,3,0,1));
      mmtmpPD1 = _mm_shufflehi_epi16(mmtmpPD1,_MM_SHUFFLE(2,3,0,1));
      mmtmpPD1 = _mm_sign_epi16(mmtmpPD1,*(__m128i*)conjugate);
      mmtmpPD1 = _mm_madd_epi16(mmtmpPD1,dl_ch128i[2]);
      // mmtmpPD1 contains imag part of 4 consecutive outputs (32-bit)
      mmtmpPD0 = _mm_srai_epi32(mmtmpPD0,output_shift);
      mmtmpPD1 = _mm_srai_epi32(mmtmpPD1,output_shift);
      mmtmpPD2 = _mm_unpacklo_epi32(mmtmpPD0,mmtmpPD1);
      mmtmpPD3 = _mm_unpackhi_epi32(mmtmpPD0,mmtmpPD1);
771

772 773 774 775
      dl_ch_rho128[2] = _mm_packs_epi32(mmtmpPD2,mmtmpPD3);
      //print_shorts("rx:",dl_ch128_2+2);
      //print_shorts("ch:",dl_ch128+2);
      //print_shorts("pack:",rho128+2);
776

777 778 779
      dl_ch128+=3;
      dl_ch128i+=3;
      dl_ch_rho128+=3;
780 781 782

    }

783
  }
784

785 786
  _mm_empty();
  _m_empty();
787 788


789 790 791 792
}


void pdcch_detection_mrc_i(LTE_DL_FRAME_PARMS *frame_parms,
793 794 795 796 797 798
                           int32_t **rxdataF_comp,
                           int32_t **rxdataF_comp_i,
                           int32_t **rho,
                           int32_t **rho_i,
                           uint8_t symbol)
{
799

800
  uint8_t aatx;
801 802

  __m128i *rxdataF_comp128_0,*rxdataF_comp128_1,*rxdataF_comp128_i0,*rxdataF_comp128_i1,*rho128_0,*rho128_1,*rho128_i0,*rho128_i1;
803
  int32_t i;
804 805

  if (frame_parms->nb_antennas_rx>1) {
806
    for (aatx=0; aatx<frame_parms->nb_antennas_tx_eNB; aatx++) {
807 808
      //if (frame_parms->mode1_flag && (aatx>0)) break;

809
      rxdataF_comp128_0   = (__m128i *)&rxdataF_comp[(aatx<<1)][symbol*frame_parms->N_RB_DL*12];
810 811 812
      rxdataF_comp128_1   = (__m128i *)&rxdataF_comp[(aatx<<1)+1][symbol*frame_parms->N_RB_DL*12];

      // MRC on each re of rb on MF output
813 814
      for (i=0; i<frame_parms->N_RB_DL*3; i++) {
        rxdataF_comp128_0[i] = _mm_adds_epi16(_mm_srai_epi16(rxdataF_comp128_0[i],1),_mm_srai_epi16(rxdataF_comp128_1[i],1));
815 816
      }
    }
817

818 819
    rho128_0 = (__m128i *) &rho[0][symbol*frame_parms->N_RB_DL*12];
    rho128_1 = (__m128i *) &rho[1][symbol*frame_parms->N_RB_DL*12];
820 821

    for (i=0; i<frame_parms->N_RB_DL*3; i++) {
822 823
      rho128_0[i] = _mm_adds_epi16(_mm_srai_epi16(rho128_0[i],1),_mm_srai_epi16(rho128_1[i],1));
    }
824

825 826
    rho128_i0 = (__m128i *) &rho_i[0][symbol*frame_parms->N_RB_DL*12];
    rho128_i1 = (__m128i *) &rho_i[1][symbol*frame_parms->N_RB_DL*12];
827
    rxdataF_comp128_i0   = (__m128i *)&rxdataF_comp_i[0][symbol*frame_parms->N_RB_DL*12];
828
    rxdataF_comp128_i1   = (__m128i *)&rxdataF_comp_i[1][symbol*frame_parms->N_RB_DL*12];
829 830 831

    // MRC on each re of rb on MF and rho
    for (i=0; i<frame_parms->N_RB_DL*3; i++) {
832 833 834 835
      rxdataF_comp128_i0[i] = _mm_adds_epi16(_mm_srai_epi16(rxdataF_comp128_i0[i],1),_mm_srai_epi16(rxdataF_comp128_i1[i],1));
      rho128_i0[i]          = _mm_adds_epi16(_mm_srai_epi16(rho128_i0[i],1),_mm_srai_epi16(rho128_i1[i],1));
    }
  }
836

837 838 839 840 841 842
  _mm_empty();
  _m_empty();

}


843
void pdcch_extract_rbs_single(int32_t **rxdataF,
844 845 846 847 848 849 850
                              int32_t **dl_ch_estimates,
                              int32_t **rxdataF_ext,
                              int32_t **dl_ch_estimates_ext,
                              uint8_t symbol,
                              uint32_t high_speed_flag,
                              LTE_DL_FRAME_PARMS *frame_parms)
{
851 852


853 854 855
  uint16_t rb,nb_rb=0;
  uint8_t i,j,aarx;
  int32_t *dl_ch0,*dl_ch0_ext,*rxF,*rxF_ext;
856

857 858

  int nushiftmod3 = frame_parms->nushift%3;
859
  uint8_t symbol_mod;
860 861 862 863 864

  symbol_mod = (symbol>=(7-frame_parms->Ncp)) ? symbol-(7-frame_parms->Ncp) : symbol;
#ifdef DEBUG_DCI_DECODING
  msg("[PHY] extract_rbs_single: symbol_mod %d\n",symbol_mod);
#endif
865 866

  for (aarx=0; aarx<frame_parms->nb_antennas_rx; aarx++) {
867 868 869 870 871

    if (high_speed_flag == 1)
      dl_ch0     = &dl_ch_estimates[aarx][5+(symbol*(frame_parms->ofdm_symbol_size))];
    else
      dl_ch0     = &dl_ch_estimates[aarx][5];
872

873 874 875 876 877
    dl_ch0_ext = &dl_ch_estimates_ext[aarx][symbol*(frame_parms->N_RB_DL*12)];

    rxF_ext   = &rxdataF_ext[aarx][symbol*(frame_parms->N_RB_DL*12)];

    rxF       = &rxdataF[aarx][(frame_parms->first_carrier_offset + (symbol*(frame_parms->ofdm_symbol_size)))];
878

879
    if ((frame_parms->N_RB_DL&1) == 0)  { // even number of RBs
880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925
      for (rb=0; rb<frame_parms->N_RB_DL; rb++) {

        // For second half of RBs skip DC carrier
        if (rb==(frame_parms->N_RB_DL>>1)) {
          rxF       = &rxdataF[aarx][(1 + (symbol*(frame_parms->ofdm_symbol_size)))];

          //dl_ch0++;
        }

        if (symbol_mod>0) {
          memcpy(dl_ch0_ext,dl_ch0,12*sizeof(int32_t));

          for (i=0; i<12; i++) {

            rxF_ext[i]=rxF[i];

          }

          nb_rb++;
          dl_ch0_ext+=12;
          rxF_ext+=12;

          dl_ch0+=12;
          rxF+=12;
        } else {
          j=0;

          for (i=0; i<12; i++) {
            if ((i!=nushiftmod3) &&
                (i!=(nushiftmod3+3)) &&
                (i!=(nushiftmod3+6)) &&
                (i!=(nushiftmod3+9))) {
              rxF_ext[j]=rxF[i];
              //                        printf("extract rb %d, re %d => (%d,%d)\n",rb,i,*(short *)&rxF_ext[j],*(1+(short*)&rxF_ext[j]));
              dl_ch0_ext[j++]=dl_ch0[i];
              //                printf("ch %d => (%d,%d)\n",i,*(short *)&dl_ch0[i],*(1+(short*)&dl_ch0[i]));
            }
          }

          nb_rb++;
          dl_ch0_ext+=8;
          rxF_ext+=8;

          dl_ch0+=12;
          rxF+=12;
        }
926
      }
927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963
    } else { // Odd number of RBs
      for (rb=0; rb<frame_parms->N_RB_DL>>1; rb++) {

        if (symbol_mod>0) {
          memcpy(dl_ch0_ext,dl_ch0,12*sizeof(int32_t));

          for (i=0; i<12; i++)
            rxF_ext[i]=rxF[i];

          nb_rb++;
          dl_ch0_ext+=12;
          rxF_ext+=12;

          dl_ch0+=12;
          rxF+=12;
        } else {
          j=0;

          for (i=0; i<12; i++) {
            if ((i!=nushiftmod3) &&
                (i!=(nushiftmod3+3)) &&
                (i!=(nushiftmod3+6)) &&
                (i!=(nushiftmod3+9))) {
              rxF_ext[j]=rxF[i];
              //                        printf("extract rb %d, re %d => (%d,%d)\n",rb,i,*(short *)&rxF_ext[j],*(1+(short*)&rxF_ext[j]));
              dl_ch0_ext[j++]=dl_ch0[i];
              //                printf("ch %d => (%d,%d)\n",i,*(short *)&dl_ch0[i],*(1+(short*)&dl_ch0[i]));
            }
          }

          nb_rb++;
          dl_ch0_ext+=8;
          rxF_ext+=8;

          dl_ch0+=12;
          rxF+=12;
        }
964
      }
965

966
      // Do middle RB (around DC)
967
      //  msg("dlch_ext %d\n",dl_ch0_ext-&dl_ch_estimates_ext[aarx][0]);
968 969

      if (symbol_mod==0) {
970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018
        j=0;

        for (i=0; i<6; i++) {
          if ((i!=nushiftmod3) &&
              (i!=(nushiftmod3+3))) {
            dl_ch0_ext[j]=dl_ch0[i];
            rxF_ext[j++]=rxF[i];
            //              printf("**extract rb %d, re %d => (%d,%d)\n",rb,i,*(short *)&rxF_ext[j-1],*(1+(short*)&rxF_ext[j-1]));
          }
        }

        rxF       = &rxdataF[aarx][((symbol*(frame_parms->ofdm_symbol_size)))];

        for (; i<12; i++) {
          if ((i!=(nushiftmod3+6)) &&
              (i!=(nushiftmod3+9))) {
            dl_ch0_ext[j]=dl_ch0[i];
            rxF_ext[j++]=rxF[(1+i-6)];
            //              printf("**extract rb %d, re %d => (%d,%d)\n",rb,i,*(short *)&rxF_ext[j-1],*(1+(short*)&rxF_ext[j-1]));
          }
        }


        nb_rb++;
        dl_ch0_ext+=8;
        rxF_ext+=8;
        dl_ch0+=12;
        rxF+=7;
        rb++;
      } else {
        for (i=0; i<6; i++) {
          dl_ch0_ext[i]=dl_ch0[i];
          rxF_ext[i]=rxF[i];
        }

        rxF       = &rxdataF[aarx][((symbol*(frame_parms->ofdm_symbol_size)))];

        for (; i<12; i++) {
          dl_ch0_ext[i]=dl_ch0[i];
          rxF_ext[i]=rxF[(1+i-6)];
        }


        nb_rb++;
        dl_ch0_ext+=12;
        rxF_ext+=12;
        dl_ch0+=12;
        rxF+=7;
        rb++;
1019 1020
      }

1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054
      for (; rb<frame_parms->N_RB_DL; rb++) {
        if (symbol_mod > 0) {
          memcpy(dl_ch0_ext,dl_ch0,12*sizeof(int32_t));

          for (i=0; i<12; i++)
            rxF_ext[i]=rxF[i];

          nb_rb++;
          dl_ch0_ext+=12;
          rxF_ext+=12;

          dl_ch0+=12;
          rxF+=12;
        } else {
          j=0;

          for (i=0; i<12; i++) {
            if ((i!=(nushiftmod3)) &&
                (i!=(nushiftmod3+3)) &&
                (i!=(nushiftmod3+6)) &&
                (i!=(nushiftmod3+9))) {
              rxF_ext[j]=rxF[i];
              //                printf("extract rb %d, re %d => (%d,%d)\n",rb,i,*(short *)&rxF_ext[j],*(1+(short*)&rxF_ext[j]));
              dl_ch0_ext[j++]=dl_ch0[i];
            }
          }

          nb_rb++;
          dl_ch0_ext+=8;
          rxF_ext+=8;

          dl_ch0+=12;
          rxF+=12;
        }
1055 1056 1057 1058 1059 1060 1061 1062 1063
      }
    }
  }

  _mm_empty();
  _m_empty();

}

1064
void pdcch_extract_rbs_dual(int32_t **rxdataF,
1065 1066 1067 1068 1069 1070 1071 1072
                            int32_t **dl_ch_estimates,
                            int32_t **rxdataF_ext,
                            int32_t **dl_ch_estimates_ext,
                            uint8_t symbol,
                            uint32_t high_speed_flag,
                            LTE_DL_FRAME_PARMS *frame_parms)
{

1073

1074 1075 1076 1077
  uint16_t rb,nb_rb=0;
  uint8_t i,aarx,j;
  int32_t *dl_ch0,*dl_ch0_ext,*dl_ch1,*dl_ch1_ext,*rxF,*rxF_ext;
  uint8_t symbol_mod;
1078 1079 1080 1081
  int nushiftmod3 = frame_parms->nushift%3;

  symbol_mod = (symbol>=(7-frame_parms->Ncp)) ? symbol-(7-frame_parms->Ncp) : symbol;

1082
  for (aarx=0; aarx<frame_parms->nb_antennas_rx; aarx++) {
1083 1084 1085 1086

    if (high_speed_flag==1) {
      dl_ch0     = &dl_ch_estimates[aarx][5+(symbol*(frame_parms->ofdm_symbol_size))];
      dl_ch1     = &dl_ch_estimates[2+aarx][5+(symbol*(frame_parms->ofdm_symbol_size))];
1087
    } else {
1088 1089 1090
      dl_ch0     = &dl_ch_estimates[aarx][5];
      dl_ch1     = &dl_ch_estimates[2+aarx][5];
    }
1091

1092 1093 1094 1095 1096 1097 1098
    dl_ch0_ext = &dl_ch_estimates_ext[aarx][symbol*(frame_parms->N_RB_DL*12)];
    dl_ch1_ext = &dl_ch_estimates_ext[2+aarx][symbol*(frame_parms->N_RB_DL*12)];

    //    msg("pdcch extract_rbs: rxF_ext pos %d\n",symbol*(frame_parms->N_RB_DL*12));
    rxF_ext   = &rxdataF_ext[aarx][symbol*(frame_parms->N_RB_DL*12)];

    rxF       = &rxdataF[aarx][(frame_parms->first_carrier_offset + (symbol*(frame_parms->ofdm_symbol_size)))];
1099

1100
    if ((frame_parms->N_RB_DL&1) == 0)  // even number of RBs
1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153
      for (rb=0; rb<frame_parms->N_RB_DL; rb++) {

        // For second half of RBs skip DC carrier
        if (rb==(frame_parms->N_RB_DL>>1)) {
          rxF       = &rxdataF[aarx][(1 + (symbol*(frame_parms->ofdm_symbol_size)))];
          //    dl_ch0++;
          //dl_ch1++;
        }

        if (symbol_mod>0) {
          memcpy(dl_ch0_ext,dl_ch0,12*sizeof(int32_t));
          memcpy(dl_ch1_ext,dl_ch1,12*sizeof(int32_t));

          /*
            msg("rb %d\n",rb);
            for (i=0;i<12;i++)
            msg("(%d %d)",((int16_t *)dl_ch0)[i<<1],((int16_t*)dl_ch0)[1+(i<<1)]);
            msg("\n");
          */
          for (i=0; i<12; i++) {
            rxF_ext[i]=rxF[i];
            //      msg("%d : (%d,%d)\n",(rxF+(2*i)-&rxdataF[aarx][( (symbol*(frame_parms->ofdm_symbol_size)))*2])/2,
            //  ((int16_t*)&rxF[i<<1])[0],((int16_t*)&rxF[i<<1])[0]);
          }

          nb_rb++;
          dl_ch0_ext+=12;
          dl_ch1_ext+=12;
          rxF_ext+=12;
        } else {
          j=0;

          for (i=0; i<12; i++) {
            if ((i!=nushiftmod3) &&
                (i!=nushiftmod3+3) &&
                (i!=nushiftmod3+6) &&
                (i!=nushiftmod3+9)) {
              rxF_ext[j]=rxF[i];
              //                            printf("extract rb %d, re %d => (%d,%d)\n",rb,i,*(short *)&rxF_ext[j],*(1+(short*)&rxF_ext[j]));
              dl_ch0_ext[j]  =dl_ch0[i];
              dl_ch1_ext[j++]=dl_ch1[i];
            }
          }

          nb_rb++;
          dl_ch0_ext+=8;
          dl_ch1_ext+=8;
          rxF_ext+=8;
        }

        dl_ch0+=12;
        dl_ch1+=12;
        rxF+=12;
1154
      }
1155

1156
    else {  // Odd number of RBs
1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205
      for (rb=0; rb<frame_parms->N_RB_DL>>1; rb++) {

        //  msg("rb %d: %d\n",rb,rxF-&rxdataF[aarx][(symbol*(frame_parms->ofdm_symbol_size))*2]);

        if (symbol_mod>0) {
          memcpy(dl_ch0_ext,dl_ch0,12*sizeof(int32_t));
          memcpy(dl_ch1_ext,dl_ch1,12*sizeof(int32_t));

          for (i=0; i<12; i++)
            rxF_ext[i]=rxF[i];

          nb_rb++;
          dl_ch0_ext+=12;
          dl_ch1_ext+=12;
          rxF_ext+=12;

          dl_ch0+=12;
          dl_ch1+=12;
          rxF+=12;

        } else {
          j=0;

          for (i=0; i<12; i++) {
            if ((i!=nushiftmod3) &&
                (i!=nushiftmod3+3) &&
                (i!=nushiftmod3+6) &&
                (i!=nushiftmod3+9)) {
              rxF_ext[j]=rxF[i];
              //                        printf("extract rb %d, re %d => (%d,%d)\n",rb,i,*(short *)&rxF_ext[j],*(1+(short*)&rxF_ext[j]));
              dl_ch0_ext[j]=dl_ch0[i];
              dl_ch1_ext[j++]=dl_ch1[i];
              //                printf("ch %d => (%d,%d)\n",i,*(short *)&dl_ch0[i],*(1+(short*)&dl_ch0[i]));
            }
          }

          nb_rb++;
          dl_ch0_ext+=8;
          dl_ch1_ext+=8;
          rxF_ext+=8;


          dl_ch0+=12;
          dl_ch1+=12;
          rxF+=12;
        }
      }

      // Do middle RB (around DC)
1206 1207

      if (symbol_mod > 0) {
1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264
        for (i=0; i<6; i++) {
          dl_ch0_ext[i]=dl_ch0[i];
          dl_ch1_ext[i]=dl_ch1[i];
          rxF_ext[i]=rxF[i];
        }

        rxF       = &rxdataF[aarx][((symbol*(frame_parms->ofdm_symbol_size)))];

        for (; i<12; i++) {
          dl_ch0_ext[i]=dl_ch0[i];
          dl_ch1_ext[i]=dl_ch1[i];
          rxF_ext[i]=rxF[(1+i)];
        }

        nb_rb++;
        dl_ch0_ext+=12;
        dl_ch1_ext+=12;
        rxF_ext+=12;

        dl_ch0+=12;
        dl_ch1+=12;
        rxF+=7;
        rb++;
      } else {
        j=0;

        for (i=0; i<6; i++) {
          if ((i!=nushiftmod3) &&
              (i!=nushiftmod3+3)) {
            dl_ch0_ext[j]=dl_ch0[i];
            dl_ch1_ext[j]=dl_ch1[i];
            rxF_ext[j++]=rxF[i];
            //              printf("**extract rb %d, re %d => (%d,%d)\n",rb,i,*(short *)&rxF_ext[j-1],*(1+(short*)&rxF_ext[j-1]));
          }
        }

        rxF       = &rxdataF[aarx][((symbol*(frame_parms->ofdm_symbol_size)))];

        for (; i<12; i++) {
          if ((i!=nushiftmod3+6) &&
              (i!=nushiftmod3+9)) {
            dl_ch0_ext[j]=dl_ch0[i];
            dl_ch1_ext[j]=dl_ch1[i];
            rxF_ext[j++]=rxF[(1+i-6)];
            //              printf("**extract rb %d, re %d => (%d,%d)\n",rb,i,*(short *)&rxF_ext[j-1],*(1+(short*)&rxF_ext[j-1]));
          }
        }


        nb_rb++;
        dl_ch0_ext+=8;
        dl_ch1_ext+=8;
        rxF_ext+=8;
        dl_ch0+=12;
        dl_ch1+=12;
        rxF+=7;
        rb++;
1265 1266
      }

1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308
      for (; rb<frame_parms->N_RB_DL; rb++) {

        if (symbol_mod>0) {
          //  msg("rb %d: %d\n",rb,rxF-&rxdataF[aarx][(symbol*(frame_parms->ofdm_symbol_size))*2]);
          memcpy(dl_ch0_ext,dl_ch0,12*sizeof(int32_t));
          memcpy(dl_ch1_ext,dl_ch1,12*sizeof(int32_t));

          for (i=0; i<12; i++)
            rxF_ext[i]=rxF[i];

          nb_rb++;
          dl_ch0_ext+=12;
          dl_ch1_ext+=12;
          rxF_ext+=12;

          dl_ch0+=12;
          dl_ch1+=12;
          rxF+=12;
        } else {
          j=0;

          for (i=0; i<12; i++) {
            if ((i!=nushiftmod3) &&
                (i!=nushiftmod3+3) &&
                (i!=nushiftmod3+6) &&
                (i!=nushiftmod3+9)) {
              rxF_ext[j]=rxF[i];
              //                printf("extract rb %d, re %d => (%d,%d)\n",rb,i,*(short *)&rxF_ext[j],*(1+(short*)&rxF_ext[j]));
              dl_ch0_ext[j]=dl_ch0[i];
              dl_ch1_ext[j++]=dl_ch1[i];
            }
          }

          nb_rb++;
          dl_ch0_ext+=8;
          dl_ch1_ext+=8;
          rxF_ext+=8;

          dl_ch0+=12;
          dl_ch1+=12;
          rxF+=12;
        }
1309 1310 1311
      }
    }
  }
1312

1313 1314
  _mm_empty();
  _m_empty();
1315

1316 1317 1318 1319

}


1320
void pdcch_channel_compensation(int32_t **rxdataF_ext,
1321 1322 1323 1324 1325 1326 1327
                                int32_t **dl_ch_estimates_ext,
                                int32_t **rxdataF_comp,
                                int32_t **rho,
                                LTE_DL_FRAME_PARMS *frame_parms,
                                uint8_t symbol,
                                uint8_t output_shift)
{
1328

1329
  uint16_t rb;
1330 1331
  __m128i *dl_ch128,*rxdataF128,*rxdataF_comp128;
  __m128i *dl_ch128_2, *rho128;
1332
  uint8_t aatx,aarx,pilots=0;
1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343




#ifdef DEBUG_DCI_DECODING
  msg("[PHY] PDCCH comp: symbol %d\n",symbol);
#endif

  if (symbol==0)
    pilots=1;

1344
  for (aatx=0; aatx<frame_parms->nb_antennas_tx_eNB; aatx++) {
1345 1346
    //if (frame_parms->mode1_flag && aatx>0) break; //if mode1_flag is set then there is only one stream to extract, independent of nb_antennas_tx_eNB

1347
    for (aarx=0; aarx<frame_parms->nb_antennas_rx; aarx++) {
1348 1349 1350 1351 1352 1353

      dl_ch128          = (__m128i *)&dl_ch_estimates_ext[(aatx<<1)+aarx][symbol*frame_parms->N_RB_DL*12];
      rxdataF128        = (__m128i *)&rxdataF_ext[aarx][symbol*frame_parms->N_RB_DL*12];
      rxdataF_comp128   = (__m128i *)&rxdataF_comp[(aatx<<1)+aarx][symbol*frame_parms->N_RB_DL*12];


1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427
      for (rb=0; rb<frame_parms->N_RB_DL; rb++) {

        // multiply by conjugated channel
        mmtmpPD0 = _mm_madd_epi16(dl_ch128[0],rxdataF128[0]);
        //  print_ints("re",&mmtmpPD0);

        // mmtmpPD0 contains real part of 4 consecutive outputs (32-bit)
        mmtmpPD1 = _mm_shufflelo_epi16(dl_ch128[0],_MM_SHUFFLE(2,3,0,1));
        mmtmpPD1 = _mm_shufflehi_epi16(mmtmpPD1,_MM_SHUFFLE(2,3,0,1));
        mmtmpPD1 = _mm_sign_epi16(mmtmpPD1,*(__m128i*)&conjugate[0]);
        //  print_ints("im",&mmtmpPD1);
        mmtmpPD1 = _mm_madd_epi16(mmtmpPD1,rxdataF128[0]);
        // mmtmpPD1 contains imag part of 4 consecutive outputs (32-bit)
        mmtmpPD0 = _mm_srai_epi32(mmtmpPD0,output_shift);
        //  print_ints("re(shift)",&mmtmpPD0);
        mmtmpPD1 = _mm_srai_epi32(mmtmpPD1,output_shift);
        //  print_ints("im(shift)",&mmtmpPD1);
        mmtmpPD2 = _mm_unpacklo_epi32(mmtmpPD0,mmtmpPD1);
        mmtmpPD3 = _mm_unpackhi_epi32(mmtmpPD0,mmtmpPD1);
        //        print_ints("c0",&mmtmpPD2);
        //  print_ints("c1",&mmtmpPD3);
        rxdataF_comp128[0] = _mm_packs_epi32(mmtmpPD2,mmtmpPD3);
        //  print_shorts("rx:",rxdataF128);
        //  print_shorts("ch:",dl_ch128);
        //  print_shorts("pack:",rxdataF_comp128);

        // multiply by conjugated channel
        mmtmpPD0 = _mm_madd_epi16(dl_ch128[1],rxdataF128[1]);
        // mmtmpPD0 contains real part of 4 consecutive outputs (32-bit)
        mmtmpPD1 = _mm_shufflelo_epi16(dl_ch128[1],_MM_SHUFFLE(2,3,0,1));
        mmtmpPD1 = _mm_shufflehi_epi16(mmtmpPD1,_MM_SHUFFLE(2,3,0,1));
        mmtmpPD1 = _mm_sign_epi16(mmtmpPD1,*(__m128i*)conjugate);
        mmtmpPD1 = _mm_madd_epi16(mmtmpPD1,rxdataF128[1]);
        // mmtmpPD1 contains imag part of 4 consecutive outputs (32-bit)
        mmtmpPD0 = _mm_srai_epi32(mmtmpPD0,output_shift);
        mmtmpPD1 = _mm_srai_epi32(mmtmpPD1,output_shift);
        mmtmpPD2 = _mm_unpacklo_epi32(mmtmpPD0,mmtmpPD1);
        mmtmpPD3 = _mm_unpackhi_epi32(mmtmpPD0,mmtmpPD1);

        rxdataF_comp128[1] = _mm_packs_epi32(mmtmpPD2,mmtmpPD3);

        //  print_shorts("rx:",rxdataF128+1);
        //  print_shorts("ch:",dl_ch128+1);
        //  print_shorts("pack:",rxdataF_comp128+1);
        // multiply by conjugated channel
        if (pilots == 0) {
          mmtmpPD0 = _mm_madd_epi16(dl_ch128[2],rxdataF128[2]);
          // mmtmpPD0 contains real part of 4 consecutive outputs (32-bit)
          mmtmpPD1 = _mm_shufflelo_epi16(dl_ch128[2],_MM_SHUFFLE(2,3,0,1));
          mmtmpPD1 = _mm_shufflehi_epi16(mmtmpPD1,_MM_SHUFFLE(2,3,0,1));
          mmtmpPD1 = _mm_sign_epi16(mmtmpPD1,*(__m128i*)conjugate);
          mmtmpPD1 = _mm_madd_epi16(mmtmpPD1,rxdataF128[2]);
          // mmtmpPD1 contains imag part of 4 consecutive outputs (32-bit)
          mmtmpPD0 = _mm_srai_epi32(mmtmpPD0,output_shift);
          mmtmpPD1 = _mm_srai_epi32(mmtmpPD1,output_shift);
          mmtmpPD2 = _mm_unpacklo_epi32(mmtmpPD0,mmtmpPD1);
          mmtmpPD3 = _mm_unpackhi_epi32(mmtmpPD0,mmtmpPD1);

          rxdataF_comp128[2] = _mm_packs_epi32(mmtmpPD2,mmtmpPD3);
        }

        //  print_shorts("rx:",rxdataF128+2);
        //  print_shorts("ch:",dl_ch128+2);
        //        print_shorts("pack:",rxdataF_comp128+2);

        if (pilots==0) {
          dl_ch128+=3;
          rxdataF128+=3;
          rxdataF_comp128+=3;
        } else {
          dl_ch128+=2;
          rxdataF128+=2;
          rxdataF_comp128+=2;
        }
1428 1429 1430
      }
    }
  }
1431

1432 1433 1434

  if (rho) {

1435
    for (aarx=0; aarx<frame_parms->nb_antennas_rx; aarx++) {
1436 1437 1438 1439
      rho128        = (__m128i *)&rho[aarx][symbol*frame_parms->N_RB_DL*12];
      dl_ch128      = (__m128i *)&dl_ch_estimates_ext[aarx][symbol*frame_parms->N_RB_DL*12];
      dl_ch128_2    = (__m128i *)&dl_ch_estimates_ext[2+aarx][symbol*frame_parms->N_RB_DL*12];

1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506
      for (rb=0; rb<frame_parms->N_RB_DL; rb++) {
        // multiply by conjugated channel
        mmtmpPD0 = _mm_madd_epi16(dl_ch128[0],dl_ch128_2[0]);
        //  print_ints("re",&mmtmpD0);

        // mmtmpD0 contains real part of 4 consecutive outputs (32-bit)
        mmtmpPD1 = _mm_shufflelo_epi16(dl_ch128[0],_MM_SHUFFLE(2,3,0,1));
        mmtmpPD1 = _mm_shufflehi_epi16(mmtmpPD1,_MM_SHUFFLE(2,3,0,1));
        mmtmpPD1 = _mm_sign_epi16(mmtmpPD1,*(__m128i*)&conjugate[0]);
        //  print_ints("im",&mmtmpPD1);
        mmtmpPD1 = _mm_madd_epi16(mmtmpPD1,dl_ch128_2[0]);
        // mmtmpPD1 contains imag part of 4 consecutive outputs (32-bit)
        mmtmpPD0 = _mm_srai_epi32(mmtmpPD0,output_shift);
        //  print_ints("re(shift)",&mmtmpD0);
        mmtmpPD1 = _mm_srai_epi32(mmtmpPD1,output_shift);
        //  print_ints("im(shift)",&mmtmpD1);
        mmtmpPD2 = _mm_unpacklo_epi32(mmtmpPD0,mmtmpPD1);
        mmtmpPD3 = _mm_unpackhi_epi32(mmtmpPD0,mmtmpPD1);
        //        print_ints("c0",&mmtmpPD2);
        //  print_ints("c1",&mmtmpPD3);
        rho128[0] = _mm_packs_epi32(mmtmpPD2,mmtmpPD3);

        //print_shorts("rx:",dl_ch128_2);
        //print_shorts("ch:",dl_ch128);
        //print_shorts("pack:",rho128);

        // multiply by conjugated channel
        mmtmpPD0 = _mm_madd_epi16(dl_ch128[1],dl_ch128_2[1]);
        // mmtmpD0 contains real part of 4 consecutive outputs (32-bit)
        mmtmpPD1 = _mm_shufflelo_epi16(dl_ch128[1],_MM_SHUFFLE(2,3,0,1));
        mmtmpPD1 = _mm_shufflehi_epi16(mmtmpPD1,_MM_SHUFFLE(2,3,0,1));
        mmtmpPD1 = _mm_sign_epi16(mmtmpPD1,*(__m128i*)conjugate);
        mmtmpPD1 = _mm_madd_epi16(mmtmpPD1,dl_ch128_2[1]);
        // mmtmpD1 contains imag part of 4 consecutive outputs (32-bit)
        mmtmpPD0 = _mm_srai_epi32(mmtmpPD0,output_shift);
        mmtmpPD1 = _mm_srai_epi32(mmtmpPD1,output_shift);
        mmtmpPD2 = _mm_unpacklo_epi32(mmtmpPD0,mmtmpPD1);
        mmtmpPD3 = _mm_unpackhi_epi32(mmtmpPD0,mmtmpPD1);


        rho128[1] =_mm_packs_epi32(mmtmpPD2,mmtmpPD3);
        //print_shorts("rx:",dl_ch128_2+1);
        //print_shorts("ch:",dl_ch128+1);
        //print_shorts("pack:",rho128+1);
        // multiply by conjugated channel
        mmtmpPD0 = _mm_madd_epi16(dl_ch128[2],dl_ch128_2[2]);
        // mmtmpPD0 contains real part of 4 consecutive outputs (32-bit)
        mmtmpPD1 = _mm_shufflelo_epi16(dl_ch128[2],_MM_SHUFFLE(2,3,0,1));
        mmtmpPD1 = _mm_shufflehi_epi16(mmtmpPD1,_MM_SHUFFLE(2,3,0,1));
        mmtmpPD1 = _mm_sign_epi16(mmtmpPD1,*(__m128i*)conjugate);
        mmtmpPD1 = _mm_madd_epi16(mmtmpPD1,dl_ch128_2[2]);
        // mmtmpPD1 contains imag part of 4 consecutive outputs (32-bit)
        mmtmpPD0 = _mm_srai_epi32(mmtmpPD0,output_shift);
        mmtmpPD1 = _mm_srai_epi32(mmtmpPD1,output_shift);
        mmtmpPD2 = _mm_unpacklo_epi32(mmtmpPD0,mmtmpPD1);
        mmtmpPD3 = _mm_unpackhi_epi32(mmtmpPD0,mmtmpPD1);

        rho128[2] = _mm_packs_epi32(mmtmpPD2,mmtmpPD3);
        //print_shorts("rx:",dl_ch128_2+2);
        //print_shorts("ch:",dl_ch128+2);
        //print_shorts("pack:",rho128+2);

        dl_ch128+=3;
        dl_ch128_2+=3;
        rho128+=3;

      }
1507 1508 1509 1510 1511 1512 1513
    }

  }

  _mm_empty();
  _m_empty();

1514
}
1515 1516

void pdcch_detection_mrc(LTE_DL_FRAME_PARMS *frame_parms,
1517 1518 1519
                         int32_t **rxdataF_comp,
                         uint8_t symbol)
{
1520

1521
  uint8_t aatx;
1522 1523

  __m128i *rxdataF_comp128_0,*rxdataF_comp128_1;
1524
  int32_t i;
1525 1526

  if (frame_parms->nb_antennas_rx>1) {
1527 1528 1529 1530
    for (aatx=0; aatx<frame_parms->nb_antennas_tx_eNB; aatx++) {
      rxdataF_comp128_0   = (__m128i *)&rxdataF_comp[(aatx<<1)][symbol*frame_parms->N_RB_DL*12];
      rxdataF_comp128_1   = (__m128i *)&rxdataF_comp[(aatx<<1)+1][symbol*frame_parms->N_RB_DL*12];

1531
      // MRC on each re of rb
1532 1533
      for (i=0; i<frame_parms->N_RB_DL*3; i++) {
        rxdataF_comp128_0[i] = _mm_adds_epi16(_mm_srai_epi16(rxdataF_comp128_0[i],1),_mm_srai_epi16(rxdataF_comp128_1[i],1));
1534 1535 1536
      }
    }
  }
gauthier's avatar