123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369 |
- /* ----------------------------------------------------------------------
- * Copyright (C) 2010-2013 ARM Limited. All rights reserved.
- *
- * $Date: 17. January 2013
- * $Revision: V1.4.1
- *
- * Project: CMSIS DSP Library
- * Title: arm_lms_q31.c
- *
- * Description: Processing function for the Q31 LMS filter.
- *
- * Target Processor: Cortex-M4/Cortex-M3/Cortex-M0
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions
- * are met:
- * - Redistributions of source code must retain the above copyright
- * notice, this list of conditions and the following disclaimer.
- * - Redistributions in binary form must reproduce the above copyright
- * notice, this list of conditions and the following disclaimer in
- * the documentation and/or other materials provided with the
- * distribution.
- * - Neither the name of ARM LIMITED nor the names of its contributors
- * may be used to endorse or promote products derived from this
- * software without specific prior written permission.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
- * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
- * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
- * FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE
- * COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
- * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
- * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
- * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
- * CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
- * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN
- * ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
- * POSSIBILITY OF SUCH DAMAGE.
- * -------------------------------------------------------------------- */
-
- #include "arm_math.h"
- /**
- * @ingroup groupFilters
- */
-
- /**
- * @addtogroup LMS
- * @{
- */
-
- /**
- * @brief Processing function for Q31 LMS filter.
- * @param[in] *S points to an instance of the Q15 LMS filter structure.
- * @param[in] *pSrc points to the block of input data.
- * @param[in] *pRef points to the block of reference data.
- * @param[out] *pOut points to the block of output data.
- * @param[out] *pErr points to the block of error data.
- * @param[in] blockSize number of samples to process.
- * @return none.
- *
- * \par Scaling and Overflow Behavior:
- * The function is implemented using an internal 64-bit accumulator.
- * The accumulator has a 2.62 format and maintains full precision of the intermediate
- * multiplication results but provides only a single guard bit.
- * Thus, if the accumulator result overflows it wraps around rather than clips.
- * In order to avoid overflows completely the input signal must be scaled down by
- * log2(numTaps) bits.
- * The reference signal should not be scaled down.
- * After all multiply-accumulates are performed, the 2.62 accumulator is shifted
- * and saturated to 1.31 format to yield the final result.
- * The output signal and error signal are in 1.31 format.
- *
- * \par
- * In this filter, filter coefficients are updated for each sample and the updation of filter cofficients are saturted.
- */
-
- void arm_lms_q31(
- const arm_lms_instance_q31 * S,
- q31_t * pSrc,
- q31_t * pRef,
- q31_t * pOut,
- q31_t * pErr,
- uint32_t blockSize)
- {
- q31_t *pState = S->pState; /* State pointer */
- uint32_t numTaps = S->numTaps; /* Number of filter coefficients in the filter */
- q31_t *pCoeffs = S->pCoeffs; /* Coefficient pointer */
- q31_t *pStateCurnt; /* Points to the current sample of the state */
- q31_t mu = S->mu; /* Adaptive factor */
- q31_t *px; /* Temporary pointer for state */
- q31_t *pb; /* Temporary pointer for coefficient buffer */
- uint32_t tapCnt, blkCnt; /* Loop counters */
- q63_t acc; /* Accumulator */
- q31_t e = 0; /* error of data sample */
- q31_t alpha; /* Intermediate constant for taps update */
- q31_t coef; /* Temporary variable for coef */
- q31_t acc_l, acc_h; /* temporary input */
- uint32_t uShift = ((uint32_t) S->postShift + 1u);
- uint32_t lShift = 32u - uShift; /* Shift to be applied to the output */
-
- /* S->pState points to buffer which contains previous frame (numTaps - 1) samples */
- /* pStateCurnt points to the location where the new input data should be written */
- pStateCurnt = &(S->pState[(numTaps - 1u)]);
-
- /* Initializing blkCnt with blockSize */
- blkCnt = blockSize;
-
-
- #ifndef ARM_MATH_CM0_FAMILY
-
- /* Run the below code for Cortex-M4 and Cortex-M3 */
-
- while(blkCnt > 0u)
- {
- /* Copy the new input sample into the state buffer */
- *pStateCurnt++ = *pSrc++;
-
- /* Initialize state pointer */
- px = pState;
-
- /* Initialize coefficient pointer */
- pb = pCoeffs;
-
- /* Set the accumulator to zero */
- acc = 0;
-
- /* Loop unrolling. Process 4 taps at a time. */
- tapCnt = numTaps >> 2;
-
- while(tapCnt > 0u)
- {
- /* Perform the multiply-accumulate */
- /* acc += b[N] * x[n-N] */
- acc += ((q63_t) (*px++)) * (*pb++);
-
- /* acc += b[N-1] * x[n-N-1] */
- acc += ((q63_t) (*px++)) * (*pb++);
-
- /* acc += b[N-2] * x[n-N-2] */
- acc += ((q63_t) (*px++)) * (*pb++);
-
- /* acc += b[N-3] * x[n-N-3] */
- acc += ((q63_t) (*px++)) * (*pb++);
-
- /* Decrement the loop counter */
- tapCnt--;
- }
-
- /* If the filter length is not a multiple of 4, compute the remaining filter taps */
- tapCnt = numTaps % 0x4u;
-
- while(tapCnt > 0u)
- {
- /* Perform the multiply-accumulate */
- acc += ((q63_t) (*px++)) * (*pb++);
-
- /* Decrement the loop counter */
- tapCnt--;
- }
-
- /* Converting the result to 1.31 format */
- /* Calc lower part of acc */
- acc_l = acc & 0xffffffff;
-
- /* Calc upper part of acc */
- acc_h = (acc >> 32) & 0xffffffff;
-
- acc = (uint32_t) acc_l >> lShift | acc_h << uShift;
-
- /* Store the result from accumulator into the destination buffer. */
- *pOut++ = (q31_t) acc;
-
- /* Compute and store error */
- e = *pRef++ - (q31_t) acc;
-
- *pErr++ = (q31_t) e;
-
- /* Compute alpha i.e. intermediate constant for taps update */
- alpha = (q31_t) (((q63_t) e * mu) >> 31);
-
- /* Initialize state pointer */
- /* Advance state pointer by 1 for the next sample */
- px = pState++;
-
- /* Initialize coefficient pointer */
- pb = pCoeffs;
-
- /* Loop unrolling. Process 4 taps at a time. */
- tapCnt = numTaps >> 2;
-
- /* Update filter coefficients */
- while(tapCnt > 0u)
- {
- /* coef is in 2.30 format */
- coef = (q31_t) (((q63_t) alpha * (*px++)) >> (32));
- /* get coef in 1.31 format by left shifting */
- *pb = clip_q63_to_q31((q63_t) * pb + (coef << 1u));
- /* update coefficient buffer to next coefficient */
- pb++;
-
- coef = (q31_t) (((q63_t) alpha * (*px++)) >> (32));
- *pb = clip_q63_to_q31((q63_t) * pb + (coef << 1u));
- pb++;
-
- coef = (q31_t) (((q63_t) alpha * (*px++)) >> (32));
- *pb = clip_q63_to_q31((q63_t) * pb + (coef << 1u));
- pb++;
-
- coef = (q31_t) (((q63_t) alpha * (*px++)) >> (32));
- *pb = clip_q63_to_q31((q63_t) * pb + (coef << 1u));
- pb++;
-
- /* Decrement the loop counter */
- tapCnt--;
- }
-
- /* If the filter length is not a multiple of 4, compute the remaining filter taps */
- tapCnt = numTaps % 0x4u;
-
- while(tapCnt > 0u)
- {
- /* Perform the multiply-accumulate */
- coef = (q31_t) (((q63_t) alpha * (*px++)) >> (32));
- *pb = clip_q63_to_q31((q63_t) * pb + (coef << 1u));
- pb++;
-
- /* Decrement the loop counter */
- tapCnt--;
- }
-
- /* Decrement the loop counter */
- blkCnt--;
- }
-
- /* Processing is complete. Now copy the last numTaps - 1 samples to the
- satrt of the state buffer. This prepares the state buffer for the
- next function call. */
-
- /* Points to the start of the pState buffer */
- pStateCurnt = S->pState;
-
- /* Loop unrolling for (numTaps - 1u) samples copy */
- tapCnt = (numTaps - 1u) >> 2u;
-
- /* copy data */
- while(tapCnt > 0u)
- {
- *pStateCurnt++ = *pState++;
- *pStateCurnt++ = *pState++;
- *pStateCurnt++ = *pState++;
- *pStateCurnt++ = *pState++;
-
- /* Decrement the loop counter */
- tapCnt--;
- }
-
- /* Calculate remaining number of copies */
- tapCnt = (numTaps - 1u) % 0x4u;
-
- /* Copy the remaining q31_t data */
- while(tapCnt > 0u)
- {
- *pStateCurnt++ = *pState++;
-
- /* Decrement the loop counter */
- tapCnt--;
- }
-
- #else
-
- /* Run the below code for Cortex-M0 */
-
- while(blkCnt > 0u)
- {
- /* Copy the new input sample into the state buffer */
- *pStateCurnt++ = *pSrc++;
-
- /* Initialize pState pointer */
- px = pState;
-
- /* Initialize pCoeffs pointer */
- pb = pCoeffs;
-
- /* Set the accumulator to zero */
- acc = 0;
-
- /* Loop over numTaps number of values */
- tapCnt = numTaps;
-
- while(tapCnt > 0u)
- {
- /* Perform the multiply-accumulate */
- acc += ((q63_t) (*px++)) * (*pb++);
-
- /* Decrement the loop counter */
- tapCnt--;
- }
-
- /* Converting the result to 1.31 format */
- /* Store the result from accumulator into the destination buffer. */
- /* Calc lower part of acc */
- acc_l = acc & 0xffffffff;
-
- /* Calc upper part of acc */
- acc_h = (acc >> 32) & 0xffffffff;
-
- acc = (uint32_t) acc_l >> lShift | acc_h << uShift;
-
- *pOut++ = (q31_t) acc;
-
- /* Compute and store error */
- e = *pRef++ - (q31_t) acc;
-
- *pErr++ = (q31_t) e;
-
- /* Weighting factor for the LMS version */
- alpha = (q31_t) (((q63_t) e * mu) >> 31);
-
- /* Initialize pState pointer */
- /* Advance state pointer by 1 for the next sample */
- px = pState++;
-
- /* Initialize pCoeffs pointer */
- pb = pCoeffs;
-
- /* Loop over numTaps number of values */
- tapCnt = numTaps;
-
- while(tapCnt > 0u)
- {
- /* Perform the multiply-accumulate */
- coef = (q31_t) (((q63_t) alpha * (*px++)) >> (32));
- *pb += (coef << 1u);
- pb++;
-
- /* Decrement the loop counter */
- tapCnt--;
- }
-
- /* Decrement the loop counter */
- blkCnt--;
- }
-
- /* Processing is complete. Now copy the last numTaps - 1 samples to the
- start of the state buffer. This prepares the state buffer for the
- next function call. */
-
- /* Points to the start of the pState buffer */
- pStateCurnt = S->pState;
-
- /* Copy (numTaps - 1u) samples */
- tapCnt = (numTaps - 1u);
-
- /* Copy the data */
- while(tapCnt > 0u)
- {
- *pStateCurnt++ = *pState++;
-
- /* Decrement the loop counter */
- tapCnt--;
- }
-
- #endif /* #ifndef ARM_MATH_CM0_FAMILY */
-
- }
-
- /**
- * @} end of LMS group
- */
|