mbed official / mbed-dsp

CMSIS DSP library

Dependents:   performance_timer Surfboard_ gps2rtty Capstone ... more

Legacy Warning

This is an mbed 2 library. To learn more about mbed OS 5, visit the docs.

cmsis_dsp/FilteringFunctions/arm_lms_norm_q31.c@1:fdd22bb7aa52, 2012-11-28 (annotated)

Committer:
emilmont
Date:
Wed Nov 28 12:30:09 2012 +0000
Revision:
1:fdd22bb7aa52
Child:
2:da51fb522205
DSP library code

Who changed what in which revision?

UserRevisionLine numberNew contents of line
emilmont 1:fdd22bb7aa52 1 /* ----------------------------------------------------------------------
emilmont 1:fdd22bb7aa52 2 * Copyright (C) 2010 ARM Limited. All rights reserved.
emilmont 1:fdd22bb7aa52 3 *
emilmont 1:fdd22bb7aa52 4 * $Date: 15. February 2012
emilmont 1:fdd22bb7aa52 5 * $Revision: V1.1.0
emilmont 1:fdd22bb7aa52 6 *
emilmont 1:fdd22bb7aa52 7 * Project: CMSIS DSP Library
emilmont 1:fdd22bb7aa52 8 * Title: arm_lms_norm_q31.c
emilmont 1:fdd22bb7aa52 9 *
emilmont 1:fdd22bb7aa52 10 * Description: Processing function for the Q31 NLMS filter.
emilmont 1:fdd22bb7aa52 11 *
emilmont 1:fdd22bb7aa52 12 * Target Processor: Cortex-M4/Cortex-M3/Cortex-M0
emilmont 1:fdd22bb7aa52 13 *
emilmont 1:fdd22bb7aa52 14 * Version 1.1.0 2012/02/15
emilmont 1:fdd22bb7aa52 15 * Updated with more optimizations, bug fixes and minor API changes.
emilmont 1:fdd22bb7aa52 16 *
emilmont 1:fdd22bb7aa52 17 * Version 1.0.10 2011/7/15
emilmont 1:fdd22bb7aa52 18 * Big Endian support added and Merged M0 and M3/M4 Source code.
emilmont 1:fdd22bb7aa52 19 *
emilmont 1:fdd22bb7aa52 20 * Version 1.0.3 2010/11/29
emilmont 1:fdd22bb7aa52 21 * Re-organized the CMSIS folders and updated documentation.
emilmont 1:fdd22bb7aa52 22 *
emilmont 1:fdd22bb7aa52 23 * Version 1.0.2 2010/11/11
emilmont 1:fdd22bb7aa52 24 * Documentation updated.
emilmont 1:fdd22bb7aa52 25 *
emilmont 1:fdd22bb7aa52 26 * Version 1.0.1 2010/10/05
emilmont 1:fdd22bb7aa52 27 * Production release and review comments incorporated.
emilmont 1:fdd22bb7aa52 28 *
emilmont 1:fdd22bb7aa52 29 * Version 1.0.0 2010/09/20
emilmont 1:fdd22bb7aa52 30 * Production release and review comments incorporated
emilmont 1:fdd22bb7aa52 31 *
emilmont 1:fdd22bb7aa52 32 * Version 0.0.7 2010/06/10
emilmont 1:fdd22bb7aa52 33 * Misra-C changes done
emilmont 1:fdd22bb7aa52 34 * -------------------------------------------------------------------- */
emilmont 1:fdd22bb7aa52 35
emilmont 1:fdd22bb7aa52 36 #include "arm_math.h"
emilmont 1:fdd22bb7aa52 37
emilmont 1:fdd22bb7aa52 38 /**
emilmont 1:fdd22bb7aa52 39 * @ingroup groupFilters
emilmont 1:fdd22bb7aa52 40 */
emilmont 1:fdd22bb7aa52 41
emilmont 1:fdd22bb7aa52 42 /**
emilmont 1:fdd22bb7aa52 43 * @addtogroup LMS_NORM
emilmont 1:fdd22bb7aa52 44 * @{
emilmont 1:fdd22bb7aa52 45 */
emilmont 1:fdd22bb7aa52 46
emilmont 1:fdd22bb7aa52 47 /**
emilmont 1:fdd22bb7aa52 48 * @brief Processing function for Q31 normalized LMS filter.
emilmont 1:fdd22bb7aa52 49 * @param[in] *S points to an instance of the Q31 normalized LMS filter structure.
emilmont 1:fdd22bb7aa52 50 * @param[in] *pSrc points to the block of input data.
emilmont 1:fdd22bb7aa52 51 * @param[in] *pRef points to the block of reference data.
emilmont 1:fdd22bb7aa52 52 * @param[out] *pOut points to the block of output data.
emilmont 1:fdd22bb7aa52 53 * @param[out] *pErr points to the block of error data.
emilmont 1:fdd22bb7aa52 54 * @param[in] blockSize number of samples to process.
emilmont 1:fdd22bb7aa52 55 * @return none.
emilmont 1:fdd22bb7aa52 56 *
emilmont 1:fdd22bb7aa52 57 * <b>Scaling and Overflow Behavior:</b>
emilmont 1:fdd22bb7aa52 58 * \par
emilmont 1:fdd22bb7aa52 59 * The function is implemented using an internal 64-bit accumulator.
emilmont 1:fdd22bb7aa52 60 * The accumulator has a 2.62 format and maintains full precision of the intermediate
emilmont 1:fdd22bb7aa52 61 * multiplication results but provides only a single guard bit.
emilmont 1:fdd22bb7aa52 62 * Thus, if the accumulator result overflows it wraps around rather than clip.
emilmont 1:fdd22bb7aa52 63 * In order to avoid overflows completely the input signal must be scaled down by
emilmont 1:fdd22bb7aa52 64 * log2(numTaps) bits. The reference signal should not be scaled down.
emilmont 1:fdd22bb7aa52 65 * After all multiply-accumulates are performed, the 2.62 accumulator is shifted
emilmont 1:fdd22bb7aa52 66 * and saturated to 1.31 format to yield the final result.
emilmont 1:fdd22bb7aa52 67 * The output signal and error signal are in 1.31 format.
emilmont 1:fdd22bb7aa52 68 *
emilmont 1:fdd22bb7aa52 69 * \par
emilmont 1:fdd22bb7aa52 70 * In this filter, filter coefficients are updated for each sample and the
emilmont 1:fdd22bb7aa52 71 * updation of filter cofficients are saturted.
emilmont 1:fdd22bb7aa52 72 *
emilmont 1:fdd22bb7aa52 73 */
emilmont 1:fdd22bb7aa52 74
emilmont 1:fdd22bb7aa52 75 void arm_lms_norm_q31(
emilmont 1:fdd22bb7aa52 76 arm_lms_norm_instance_q31 * S,
emilmont 1:fdd22bb7aa52 77 q31_t * pSrc,
emilmont 1:fdd22bb7aa52 78 q31_t * pRef,
emilmont 1:fdd22bb7aa52 79 q31_t * pOut,
emilmont 1:fdd22bb7aa52 80 q31_t * pErr,
emilmont 1:fdd22bb7aa52 81 uint32_t blockSize)
emilmont 1:fdd22bb7aa52 82 {
emilmont 1:fdd22bb7aa52 83 q31_t *pState = S->pState; /* State pointer */
emilmont 1:fdd22bb7aa52 84 q31_t *pCoeffs = S->pCoeffs; /* Coefficient pointer */
emilmont 1:fdd22bb7aa52 85 q31_t *pStateCurnt; /* Points to the current sample of the state */
emilmont 1:fdd22bb7aa52 86 q31_t *px, *pb; /* Temporary pointers for state and coefficient buffers */
emilmont 1:fdd22bb7aa52 87 q31_t mu = S->mu; /* Adaptive factor */
emilmont 1:fdd22bb7aa52 88 uint32_t numTaps = S->numTaps; /* Number of filter coefficients in the filter */
emilmont 1:fdd22bb7aa52 89 uint32_t tapCnt, blkCnt; /* Loop counters */
emilmont 1:fdd22bb7aa52 90 q63_t energy; /* Energy of the input */
emilmont 1:fdd22bb7aa52 91 q63_t acc; /* Accumulator */
emilmont 1:fdd22bb7aa52 92 q31_t e = 0, d = 0; /* error, reference data sample */
emilmont 1:fdd22bb7aa52 93 q31_t w = 0, in; /* weight factor and state */
emilmont 1:fdd22bb7aa52 94 q31_t x0; /* temporary variable to hold input sample */
emilmont 1:fdd22bb7aa52 95 // uint32_t shift = 32u - ((uint32_t) S->postShift + 1u); /* Shift to be applied to the output */
emilmont 1:fdd22bb7aa52 96 q31_t errorXmu, oneByEnergy; /* Temporary variables to store error and mu product and reciprocal of energy */
emilmont 1:fdd22bb7aa52 97 q31_t postShift; /* Post shift to be applied to weight after reciprocal calculation */
emilmont 1:fdd22bb7aa52 98 q31_t coef; /* Temporary variable for coef */
emilmont 1:fdd22bb7aa52 99 q31_t acc_l, acc_h; /* temporary input */
emilmont 1:fdd22bb7aa52 100 uint32_t uShift = ((uint32_t) S->postShift + 1u);
emilmont 1:fdd22bb7aa52 101 uint32_t lShift = 32u - uShift; /* Shift to be applied to the output */
emilmont 1:fdd22bb7aa52 102
emilmont 1:fdd22bb7aa52 103 energy = S->energy;
emilmont 1:fdd22bb7aa52 104 x0 = S->x0;
emilmont 1:fdd22bb7aa52 105
emilmont 1:fdd22bb7aa52 106 /* S->pState points to buffer which contains previous frame (numTaps - 1) samples */
emilmont 1:fdd22bb7aa52 107 /* pStateCurnt points to the location where the new input data should be written */
emilmont 1:fdd22bb7aa52 108 pStateCurnt = &(S->pState[(numTaps - 1u)]);
emilmont 1:fdd22bb7aa52 109
emilmont 1:fdd22bb7aa52 110 /* Loop over blockSize number of values */
emilmont 1:fdd22bb7aa52 111 blkCnt = blockSize;
emilmont 1:fdd22bb7aa52 112
emilmont 1:fdd22bb7aa52 113
emilmont 1:fdd22bb7aa52 114 #ifndef ARM_MATH_CM0
emilmont 1:fdd22bb7aa52 115
emilmont 1:fdd22bb7aa52 116 /* Run the below code for Cortex-M4 and Cortex-M3 */
emilmont 1:fdd22bb7aa52 117
emilmont 1:fdd22bb7aa52 118 while(blkCnt > 0u)
emilmont 1:fdd22bb7aa52 119 {
emilmont 1:fdd22bb7aa52 120
emilmont 1:fdd22bb7aa52 121 /* Copy the new input sample into the state buffer */
emilmont 1:fdd22bb7aa52 122 *pStateCurnt++ = *pSrc;
emilmont 1:fdd22bb7aa52 123
emilmont 1:fdd22bb7aa52 124 /* Initialize pState pointer */
emilmont 1:fdd22bb7aa52 125 px = pState;
emilmont 1:fdd22bb7aa52 126
emilmont 1:fdd22bb7aa52 127 /* Initialize coeff pointer */
emilmont 1:fdd22bb7aa52 128 pb = (pCoeffs);
emilmont 1:fdd22bb7aa52 129
emilmont 1:fdd22bb7aa52 130 /* Read the sample from input buffer */
emilmont 1:fdd22bb7aa52 131 in = *pSrc++;
emilmont 1:fdd22bb7aa52 132
emilmont 1:fdd22bb7aa52 133 /* Update the energy calculation */
emilmont 1:fdd22bb7aa52 134 energy = (q31_t) ((((q63_t) energy << 32) -
emilmont 1:fdd22bb7aa52 135 (((q63_t) x0 * x0) << 1)) >> 32);
emilmont 1:fdd22bb7aa52 136 energy = (q31_t) (((((q63_t) in * in) << 1) + (energy << 32)) >> 32);
emilmont 1:fdd22bb7aa52 137
emilmont 1:fdd22bb7aa52 138 /* Set the accumulator to zero */
emilmont 1:fdd22bb7aa52 139 acc = 0;
emilmont 1:fdd22bb7aa52 140
emilmont 1:fdd22bb7aa52 141 /* Loop unrolling. Process 4 taps at a time. */
emilmont 1:fdd22bb7aa52 142 tapCnt = numTaps >> 2;
emilmont 1:fdd22bb7aa52 143
emilmont 1:fdd22bb7aa52 144 while(tapCnt > 0u)
emilmont 1:fdd22bb7aa52 145 {
emilmont 1:fdd22bb7aa52 146 /* Perform the multiply-accumulate */
emilmont 1:fdd22bb7aa52 147 acc += ((q63_t) (*px++)) * (*pb++);
emilmont 1:fdd22bb7aa52 148 acc += ((q63_t) (*px++)) * (*pb++);
emilmont 1:fdd22bb7aa52 149 acc += ((q63_t) (*px++)) * (*pb++);
emilmont 1:fdd22bb7aa52 150 acc += ((q63_t) (*px++)) * (*pb++);
emilmont 1:fdd22bb7aa52 151
emilmont 1:fdd22bb7aa52 152 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 153 tapCnt--;
emilmont 1:fdd22bb7aa52 154 }
emilmont 1:fdd22bb7aa52 155
emilmont 1:fdd22bb7aa52 156 /* If the filter length is not a multiple of 4, compute the remaining filter taps */
emilmont 1:fdd22bb7aa52 157 tapCnt = numTaps % 0x4u;
emilmont 1:fdd22bb7aa52 158
emilmont 1:fdd22bb7aa52 159 while(tapCnt > 0u)
emilmont 1:fdd22bb7aa52 160 {
emilmont 1:fdd22bb7aa52 161 /* Perform the multiply-accumulate */
emilmont 1:fdd22bb7aa52 162 acc += ((q63_t) (*px++)) * (*pb++);
emilmont 1:fdd22bb7aa52 163
emilmont 1:fdd22bb7aa52 164 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 165 tapCnt--;
emilmont 1:fdd22bb7aa52 166 }
emilmont 1:fdd22bb7aa52 167
emilmont 1:fdd22bb7aa52 168 /* Converting the result to 1.31 format */
emilmont 1:fdd22bb7aa52 169 /* Calc lower part of acc */
emilmont 1:fdd22bb7aa52 170 acc_l = acc & 0xffffffff;
emilmont 1:fdd22bb7aa52 171
emilmont 1:fdd22bb7aa52 172 /* Calc upper part of acc */
emilmont 1:fdd22bb7aa52 173 acc_h = (acc >> 32) & 0xffffffff;
emilmont 1:fdd22bb7aa52 174
emilmont 1:fdd22bb7aa52 175 acc = (uint32_t) acc_l >> lShift | acc_h << uShift;
emilmont 1:fdd22bb7aa52 176
emilmont 1:fdd22bb7aa52 177 /* Store the result from accumulator into the destination buffer. */
emilmont 1:fdd22bb7aa52 178 *pOut++ = (q31_t) acc;
emilmont 1:fdd22bb7aa52 179
emilmont 1:fdd22bb7aa52 180 /* Compute and store error */
emilmont 1:fdd22bb7aa52 181 d = *pRef++;
emilmont 1:fdd22bb7aa52 182 e = d - (q31_t) acc;
emilmont 1:fdd22bb7aa52 183 *pErr++ = e;
emilmont 1:fdd22bb7aa52 184
emilmont 1:fdd22bb7aa52 185 /* Calculates the reciprocal of energy */
emilmont 1:fdd22bb7aa52 186 postShift = arm_recip_q31(energy + DELTA_Q31,
emilmont 1:fdd22bb7aa52 187 &oneByEnergy, &S->recipTable[0]);
emilmont 1:fdd22bb7aa52 188
emilmont 1:fdd22bb7aa52 189 /* Calculation of product of (e * mu) */
emilmont 1:fdd22bb7aa52 190 errorXmu = (q31_t) (((q63_t) e * mu) >> 31);
emilmont 1:fdd22bb7aa52 191
emilmont 1:fdd22bb7aa52 192 /* Weighting factor for the normalized version */
emilmont 1:fdd22bb7aa52 193 w = clip_q63_to_q31(((q63_t) errorXmu * oneByEnergy) >> (31 - postShift));
emilmont 1:fdd22bb7aa52 194
emilmont 1:fdd22bb7aa52 195 /* Initialize pState pointer */
emilmont 1:fdd22bb7aa52 196 px = pState;
emilmont 1:fdd22bb7aa52 197
emilmont 1:fdd22bb7aa52 198 /* Initialize coeff pointer */
emilmont 1:fdd22bb7aa52 199 pb = (pCoeffs);
emilmont 1:fdd22bb7aa52 200
emilmont 1:fdd22bb7aa52 201 /* Loop unrolling. Process 4 taps at a time. */
emilmont 1:fdd22bb7aa52 202 tapCnt = numTaps >> 2;
emilmont 1:fdd22bb7aa52 203
emilmont 1:fdd22bb7aa52 204 /* Update filter coefficients */
emilmont 1:fdd22bb7aa52 205 while(tapCnt > 0u)
emilmont 1:fdd22bb7aa52 206 {
emilmont 1:fdd22bb7aa52 207 /* Perform the multiply-accumulate */
emilmont 1:fdd22bb7aa52 208
emilmont 1:fdd22bb7aa52 209 /* coef is in 2.30 format */
emilmont 1:fdd22bb7aa52 210 coef = (q31_t) (((q63_t) w * (*px++)) >> (32));
emilmont 1:fdd22bb7aa52 211 /* get coef in 1.31 format by left shifting */
emilmont 1:fdd22bb7aa52 212 *pb = clip_q63_to_q31((q63_t) * pb + (coef << 1u));
emilmont 1:fdd22bb7aa52 213 /* update coefficient buffer to next coefficient */
emilmont 1:fdd22bb7aa52 214 pb++;
emilmont 1:fdd22bb7aa52 215
emilmont 1:fdd22bb7aa52 216 coef = (q31_t) (((q63_t) w * (*px++)) >> (32));
emilmont 1:fdd22bb7aa52 217 *pb = clip_q63_to_q31((q63_t) * pb + (coef << 1u));
emilmont 1:fdd22bb7aa52 218 pb++;
emilmont 1:fdd22bb7aa52 219
emilmont 1:fdd22bb7aa52 220 coef = (q31_t) (((q63_t) w * (*px++)) >> (32));
emilmont 1:fdd22bb7aa52 221 *pb = clip_q63_to_q31((q63_t) * pb + (coef << 1u));
emilmont 1:fdd22bb7aa52 222 pb++;
emilmont 1:fdd22bb7aa52 223
emilmont 1:fdd22bb7aa52 224 coef = (q31_t) (((q63_t) w * (*px++)) >> (32));
emilmont 1:fdd22bb7aa52 225 *pb = clip_q63_to_q31((q63_t) * pb + (coef << 1u));
emilmont 1:fdd22bb7aa52 226 pb++;
emilmont 1:fdd22bb7aa52 227
emilmont 1:fdd22bb7aa52 228 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 229 tapCnt--;
emilmont 1:fdd22bb7aa52 230 }
emilmont 1:fdd22bb7aa52 231
emilmont 1:fdd22bb7aa52 232 /* If the filter length is not a multiple of 4, compute the remaining filter taps */
emilmont 1:fdd22bb7aa52 233 tapCnt = numTaps % 0x4u;
emilmont 1:fdd22bb7aa52 234
emilmont 1:fdd22bb7aa52 235 while(tapCnt > 0u)
emilmont 1:fdd22bb7aa52 236 {
emilmont 1:fdd22bb7aa52 237 /* Perform the multiply-accumulate */
emilmont 1:fdd22bb7aa52 238 coef = (q31_t) (((q63_t) w * (*px++)) >> (32));
emilmont 1:fdd22bb7aa52 239 *pb = clip_q63_to_q31((q63_t) * pb + (coef << 1u));
emilmont 1:fdd22bb7aa52 240 pb++;
emilmont 1:fdd22bb7aa52 241
emilmont 1:fdd22bb7aa52 242 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 243 tapCnt--;
emilmont 1:fdd22bb7aa52 244 }
emilmont 1:fdd22bb7aa52 245
emilmont 1:fdd22bb7aa52 246 /* Read the sample from state buffer */
emilmont 1:fdd22bb7aa52 247 x0 = *pState;
emilmont 1:fdd22bb7aa52 248
emilmont 1:fdd22bb7aa52 249 /* Advance state pointer by 1 for the next sample */
emilmont 1:fdd22bb7aa52 250 pState = pState + 1;
emilmont 1:fdd22bb7aa52 251
emilmont 1:fdd22bb7aa52 252 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 253 blkCnt--;
emilmont 1:fdd22bb7aa52 254 }
emilmont 1:fdd22bb7aa52 255
emilmont 1:fdd22bb7aa52 256 /* Save energy and x0 values for the next frame */
emilmont 1:fdd22bb7aa52 257 S->energy = (q31_t) energy;
emilmont 1:fdd22bb7aa52 258 S->x0 = x0;
emilmont 1:fdd22bb7aa52 259
emilmont 1:fdd22bb7aa52 260 /* Processing is complete. Now copy the last numTaps - 1 samples to the
emilmont 1:fdd22bb7aa52 261 satrt of the state buffer. This prepares the state buffer for the
emilmont 1:fdd22bb7aa52 262 next function call. */
emilmont 1:fdd22bb7aa52 263
emilmont 1:fdd22bb7aa52 264 /* Points to the start of the pState buffer */
emilmont 1:fdd22bb7aa52 265 pStateCurnt = S->pState;
emilmont 1:fdd22bb7aa52 266
emilmont 1:fdd22bb7aa52 267 /* Loop unrolling for (numTaps - 1u) samples copy */
emilmont 1:fdd22bb7aa52 268 tapCnt = (numTaps - 1u) >> 2u;
emilmont 1:fdd22bb7aa52 269
emilmont 1:fdd22bb7aa52 270 /* copy data */
emilmont 1:fdd22bb7aa52 271 while(tapCnt > 0u)
emilmont 1:fdd22bb7aa52 272 {
emilmont 1:fdd22bb7aa52 273 *pStateCurnt++ = *pState++;
emilmont 1:fdd22bb7aa52 274 *pStateCurnt++ = *pState++;
emilmont 1:fdd22bb7aa52 275 *pStateCurnt++ = *pState++;
emilmont 1:fdd22bb7aa52 276 *pStateCurnt++ = *pState++;
emilmont 1:fdd22bb7aa52 277
emilmont 1:fdd22bb7aa52 278 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 279 tapCnt--;
emilmont 1:fdd22bb7aa52 280 }
emilmont 1:fdd22bb7aa52 281
emilmont 1:fdd22bb7aa52 282 /* Calculate remaining number of copies */
emilmont 1:fdd22bb7aa52 283 tapCnt = (numTaps - 1u) % 0x4u;
emilmont 1:fdd22bb7aa52 284
emilmont 1:fdd22bb7aa52 285 /* Copy the remaining q31_t data */
emilmont 1:fdd22bb7aa52 286 while(tapCnt > 0u)
emilmont 1:fdd22bb7aa52 287 {
emilmont 1:fdd22bb7aa52 288 *pStateCurnt++ = *pState++;
emilmont 1:fdd22bb7aa52 289
emilmont 1:fdd22bb7aa52 290 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 291 tapCnt--;
emilmont 1:fdd22bb7aa52 292 }
emilmont 1:fdd22bb7aa52 293
emilmont 1:fdd22bb7aa52 294 #else
emilmont 1:fdd22bb7aa52 295
emilmont 1:fdd22bb7aa52 296 /* Run the below code for Cortex-M0 */
emilmont 1:fdd22bb7aa52 297
emilmont 1:fdd22bb7aa52 298 while(blkCnt > 0u)
emilmont 1:fdd22bb7aa52 299 {
emilmont 1:fdd22bb7aa52 300
emilmont 1:fdd22bb7aa52 301 /* Copy the new input sample into the state buffer */
emilmont 1:fdd22bb7aa52 302 *pStateCurnt++ = *pSrc;
emilmont 1:fdd22bb7aa52 303
emilmont 1:fdd22bb7aa52 304 /* Initialize pState pointer */
emilmont 1:fdd22bb7aa52 305 px = pState;
emilmont 1:fdd22bb7aa52 306
emilmont 1:fdd22bb7aa52 307 /* Initialize pCoeffs pointer */
emilmont 1:fdd22bb7aa52 308 pb = pCoeffs;
emilmont 1:fdd22bb7aa52 309
emilmont 1:fdd22bb7aa52 310 /* Read the sample from input buffer */
emilmont 1:fdd22bb7aa52 311 in = *pSrc++;
emilmont 1:fdd22bb7aa52 312
emilmont 1:fdd22bb7aa52 313 /* Update the energy calculation */
emilmont 1:fdd22bb7aa52 314 energy =
emilmont 1:fdd22bb7aa52 315 (q31_t) ((((q63_t) energy << 32) - (((q63_t) x0 * x0) << 1)) >> 32);
emilmont 1:fdd22bb7aa52 316 energy = (q31_t) (((((q63_t) in * in) << 1) + (energy << 32)) >> 32);
emilmont 1:fdd22bb7aa52 317
emilmont 1:fdd22bb7aa52 318 /* Set the accumulator to zero */
emilmont 1:fdd22bb7aa52 319 acc = 0;
emilmont 1:fdd22bb7aa52 320
emilmont 1:fdd22bb7aa52 321 /* Loop over numTaps number of values */
emilmont 1:fdd22bb7aa52 322 tapCnt = numTaps;
emilmont 1:fdd22bb7aa52 323
emilmont 1:fdd22bb7aa52 324 while(tapCnt > 0u)
emilmont 1:fdd22bb7aa52 325 {
emilmont 1:fdd22bb7aa52 326 /* Perform the multiply-accumulate */
emilmont 1:fdd22bb7aa52 327 acc += ((q63_t) (*px++)) * (*pb++);
emilmont 1:fdd22bb7aa52 328
emilmont 1:fdd22bb7aa52 329 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 330 tapCnt--;
emilmont 1:fdd22bb7aa52 331 }
emilmont 1:fdd22bb7aa52 332
emilmont 1:fdd22bb7aa52 333 /* Converting the result to 1.31 format */
emilmont 1:fdd22bb7aa52 334 /* Converting the result to 1.31 format */
emilmont 1:fdd22bb7aa52 335 /* Calc lower part of acc */
emilmont 1:fdd22bb7aa52 336 acc_l = acc & 0xffffffff;
emilmont 1:fdd22bb7aa52 337
emilmont 1:fdd22bb7aa52 338 /* Calc upper part of acc */
emilmont 1:fdd22bb7aa52 339 acc_h = (acc >> 32) & 0xffffffff;
emilmont 1:fdd22bb7aa52 340
emilmont 1:fdd22bb7aa52 341 acc = (uint32_t) acc_l >> lShift | acc_h << uShift;
emilmont 1:fdd22bb7aa52 342
emilmont 1:fdd22bb7aa52 343
emilmont 1:fdd22bb7aa52 344 //acc = (q31_t) (acc >> shift);
emilmont 1:fdd22bb7aa52 345
emilmont 1:fdd22bb7aa52 346 /* Store the result from accumulator into the destination buffer. */
emilmont 1:fdd22bb7aa52 347 *pOut++ = (q31_t) acc;
emilmont 1:fdd22bb7aa52 348
emilmont 1:fdd22bb7aa52 349 /* Compute and store error */
emilmont 1:fdd22bb7aa52 350 d = *pRef++;
emilmont 1:fdd22bb7aa52 351 e = d - (q31_t) acc;
emilmont 1:fdd22bb7aa52 352 *pErr++ = e;
emilmont 1:fdd22bb7aa52 353
emilmont 1:fdd22bb7aa52 354 /* Calculates the reciprocal of energy */
emilmont 1:fdd22bb7aa52 355 postShift =
emilmont 1:fdd22bb7aa52 356 arm_recip_q31(energy + DELTA_Q31, &oneByEnergy, &S->recipTable[0]);
emilmont 1:fdd22bb7aa52 357
emilmont 1:fdd22bb7aa52 358 /* Calculation of product of (e * mu) */
emilmont 1:fdd22bb7aa52 359 errorXmu = (q31_t) (((q63_t) e * mu) >> 31);
emilmont 1:fdd22bb7aa52 360
emilmont 1:fdd22bb7aa52 361 /* Weighting factor for the normalized version */
emilmont 1:fdd22bb7aa52 362 w = clip_q63_to_q31(((q63_t) errorXmu * oneByEnergy) >> (31 - postShift));
emilmont 1:fdd22bb7aa52 363
emilmont 1:fdd22bb7aa52 364 /* Initialize pState pointer */
emilmont 1:fdd22bb7aa52 365 px = pState;
emilmont 1:fdd22bb7aa52 366
emilmont 1:fdd22bb7aa52 367 /* Initialize coeff pointer */
emilmont 1:fdd22bb7aa52 368 pb = (pCoeffs);
emilmont 1:fdd22bb7aa52 369
emilmont 1:fdd22bb7aa52 370 /* Loop over numTaps number of values */
emilmont 1:fdd22bb7aa52 371 tapCnt = numTaps;
emilmont 1:fdd22bb7aa52 372
emilmont 1:fdd22bb7aa52 373 while(tapCnt > 0u)
emilmont 1:fdd22bb7aa52 374 {
emilmont 1:fdd22bb7aa52 375 /* Perform the multiply-accumulate */
emilmont 1:fdd22bb7aa52 376 /* coef is in 2.30 format */
emilmont 1:fdd22bb7aa52 377 coef = (q31_t) (((q63_t) w * (*px++)) >> (32));
emilmont 1:fdd22bb7aa52 378 /* get coef in 1.31 format by left shifting */
emilmont 1:fdd22bb7aa52 379 *pb = clip_q63_to_q31((q63_t) * pb + (coef << 1u));
emilmont 1:fdd22bb7aa52 380 /* update coefficient buffer to next coefficient */
emilmont 1:fdd22bb7aa52 381 pb++;
emilmont 1:fdd22bb7aa52 382
emilmont 1:fdd22bb7aa52 383 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 384 tapCnt--;
emilmont 1:fdd22bb7aa52 385 }
emilmont 1:fdd22bb7aa52 386
emilmont 1:fdd22bb7aa52 387 /* Read the sample from state buffer */
emilmont 1:fdd22bb7aa52 388 x0 = *pState;
emilmont 1:fdd22bb7aa52 389
emilmont 1:fdd22bb7aa52 390 /* Advance state pointer by 1 for the next sample */
emilmont 1:fdd22bb7aa52 391 pState = pState + 1;
emilmont 1:fdd22bb7aa52 392
emilmont 1:fdd22bb7aa52 393 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 394 blkCnt--;
emilmont 1:fdd22bb7aa52 395 }
emilmont 1:fdd22bb7aa52 396
emilmont 1:fdd22bb7aa52 397 /* Save energy and x0 values for the next frame */
emilmont 1:fdd22bb7aa52 398 S->energy = (q31_t) energy;
emilmont 1:fdd22bb7aa52 399 S->x0 = x0;
emilmont 1:fdd22bb7aa52 400
emilmont 1:fdd22bb7aa52 401 /* Processing is complete. Now copy the last numTaps - 1 samples to the
emilmont 1:fdd22bb7aa52 402 start of the state buffer. This prepares the state buffer for the
emilmont 1:fdd22bb7aa52 403 next function call. */
emilmont 1:fdd22bb7aa52 404
emilmont 1:fdd22bb7aa52 405 /* Points to the start of the pState buffer */
emilmont 1:fdd22bb7aa52 406 pStateCurnt = S->pState;
emilmont 1:fdd22bb7aa52 407
emilmont 1:fdd22bb7aa52 408 /* Loop for (numTaps - 1u) samples copy */
emilmont 1:fdd22bb7aa52 409 tapCnt = (numTaps - 1u);
emilmont 1:fdd22bb7aa52 410
emilmont 1:fdd22bb7aa52 411 /* Copy the remaining q31_t data */
emilmont 1:fdd22bb7aa52 412 while(tapCnt > 0u)
emilmont 1:fdd22bb7aa52 413 {
emilmont 1:fdd22bb7aa52 414 *pStateCurnt++ = *pState++;
emilmont 1:fdd22bb7aa52 415
emilmont 1:fdd22bb7aa52 416 /* Decrement the loop counter */
emilmont 1:fdd22bb7aa52 417 tapCnt--;
emilmont 1:fdd22bb7aa52 418 }
emilmont 1:fdd22bb7aa52 419
emilmont 1:fdd22bb7aa52 420 #endif /* #ifndef ARM_MATH_CM0 */
emilmont 1:fdd22bb7aa52 421
emilmont 1:fdd22bb7aa52 422 }
emilmont 1:fdd22bb7aa52 423
emilmont 1:fdd22bb7aa52 424 /**
emilmont 1:fdd22bb7aa52 425 * @} end of LMS_NORM group
emilmont 1:fdd22bb7aa52 426 */