dsp - CMSIS DSP Library from CMSIS 2.0. See http://www.…

Users » simon » Code » dsp

CMSIS DSP Library from CMSIS 2.0. See http://www.onarm.com/cmsis/ for full details

Dependents: K22F_DSP_Matrix_least_square BNO055-ELEC3810 1BNO055 ECE4180Project--Slave2 ... more

src/Cortex-M4-M3/FilteringFunctions/arm_lms_q31.c@0:1014af42efd9, 2011-03-10 (annotated)

Committer:: simon
Date:: Thu Mar 10 15:07:50 2011 +0000
Revision:: 0:1014af42efd9

Who changed what in which revision?

User	Revision	Line number	New contents of line
simon	0:1014af42efd9	1	/* ----------------------------------------------------------------------
simon	0:1014af42efd9	2	* Copyright (C) 2010 ARM Limited. All rights reserved.
simon	0:1014af42efd9	3	*
simon	0:1014af42efd9	4	* $Date: 29. November 2010
simon	0:1014af42efd9	5	* $Revision: V1.0.3
simon	0:1014af42efd9	6	*
simon	0:1014af42efd9	7	* Project: CMSIS DSP Library
simon	0:1014af42efd9	8	* Title: arm_lms_q31.c
simon	0:1014af42efd9	9	*
simon	0:1014af42efd9	10	* Description: Processing function for the Q31 LMS filter.
simon	0:1014af42efd9	11	*
simon	0:1014af42efd9	12	* Target Processor: Cortex-M4/Cortex-M3
simon	0:1014af42efd9	13	*
simon	0:1014af42efd9	14	* Version 1.0.3 2010/11/29
simon	0:1014af42efd9	15	* Re-organized the CMSIS folders and updated documentation.
simon	0:1014af42efd9	16	*
simon	0:1014af42efd9	17	* Version 1.0.2 2010/11/11
simon	0:1014af42efd9	18	* Documentation updated.
simon	0:1014af42efd9	19	*
simon	0:1014af42efd9	20	* Version 1.0.1 2010/10/05
simon	0:1014af42efd9	21	* Production release and review comments incorporated.
simon	0:1014af42efd9	22	*
simon	0:1014af42efd9	23	* Version 1.0.0 2010/09/20
simon	0:1014af42efd9	24	* Production release and review comments incorporated
simon	0:1014af42efd9	25	*
simon	0:1014af42efd9	26	* Version 0.0.7 2010/06/10
simon	0:1014af42efd9	27	* Misra-C changes done
simon	0:1014af42efd9	28	* -------------------------------------------------------------------- */
simon	0:1014af42efd9	29
simon	0:1014af42efd9	30	#include "arm_math.h"
simon	0:1014af42efd9	31	/**
simon	0:1014af42efd9	32	* @ingroup groupFilters
simon	0:1014af42efd9	33	*/
simon	0:1014af42efd9	34
simon	0:1014af42efd9	35	/**
simon	0:1014af42efd9	36	* @addtogroup LMS
simon	0:1014af42efd9	37	* @{
simon	0:1014af42efd9	38	*/
simon	0:1014af42efd9	39
simon	0:1014af42efd9	40	/**
simon	0:1014af42efd9	41	* @brief Processing function for Q31 LMS filter.
simon	0:1014af42efd9	42	* @param[in] *S points to an instance of the Q15 LMS filter structure.
simon	0:1014af42efd9	43	* @param[in] *pSrc points to the block of input data.
simon	0:1014af42efd9	44	* @param[in] *pRef points to the block of reference data.
simon	0:1014af42efd9	45	* @param[out] *pOut points to the block of output data.
simon	0:1014af42efd9	46	* @param[out] *pErr points to the block of error data.
simon	0:1014af42efd9	47	* @param[in] blockSize number of samples to process.
simon	0:1014af42efd9	48	* @return none.
simon	0:1014af42efd9	49	*
simon	0:1014af42efd9	50	* \par Scaling and Overflow Behavior:
simon	0:1014af42efd9	51	* The function is implemented using an internal 64-bit accumulator.
simon	0:1014af42efd9	52	* The accumulator has a 2.62 format and maintains full precision of the intermediate multiplication results but provides only a single guard bit.
simon	0:1014af42efd9	53	* Thus, if the accumulator result overflows it wraps around rather than clips.
simon	0:1014af42efd9	54	* In order to avoid overflows completely the input signal must be scaled down by log2(numTaps) bits.
simon	0:1014af42efd9	55	* The reference signal should not be scaled down.
simon	0:1014af42efd9	56	* After all multiply-accumulates are performed, the 2.62 accumulator is truncated to 1.32 format and then saturated to 1.31 format.
simon	0:1014af42efd9	57	* The output signal and error signal are in 1.31 format.
simon	0:1014af42efd9	58	*
simon	0:1014af42efd9	59	* \par
simon	0:1014af42efd9	60	* In this filter, filter coefficients are updated for each sample and the updation of filter cofficients are saturted.
simon	0:1014af42efd9	61	*/
simon	0:1014af42efd9	62
simon	0:1014af42efd9	63	void arm_lms_q31(
simon	0:1014af42efd9	64	const arm_lms_instance_q31 * S,
simon	0:1014af42efd9	65	q31_t * pSrc,
simon	0:1014af42efd9	66	q31_t * pRef,
simon	0:1014af42efd9	67	q31_t * pOut,
simon	0:1014af42efd9	68	q31_t * pErr,
simon	0:1014af42efd9	69	uint32_t blockSize)
simon	0:1014af42efd9	70	{
simon	0:1014af42efd9	71	q31_t pState = S->pState; / State pointer */
simon	0:1014af42efd9	72	uint32_t numTaps = S->numTaps; /* Number of filter coefficients in the filter */
simon	0:1014af42efd9	73	q31_t pCoeffs = S->pCoeffs; / Coefficient pointer */
simon	0:1014af42efd9	74	q31_t pStateCurnt; / Points to the current sample of the state */
simon	0:1014af42efd9	75	q31_t mu = S->mu; /* Adaptive factor */
simon	0:1014af42efd9	76	q31_t px; / Temporary pointer for state */
simon	0:1014af42efd9	77	q31_t pb; / Temporary pointer for coefficient buffer */
simon	0:1014af42efd9	78	uint32_t tapCnt, blkCnt; /* Loop counters */
simon	0:1014af42efd9	79	q63_t acc; /* Accumulator */
simon	0:1014af42efd9	80	q31_t e = 0; /* error of data sample */
simon	0:1014af42efd9	81	q31_t alpha; /* Intermediate constant for taps update */
simon	0:1014af42efd9	82	uint8_t shift = (uint8_t) (32u - (S->postShift + 1u)); /* Shift to be applied to the output */
simon	0:1014af42efd9	83	q31_t coef; /* Temporary variable for coef */
simon	0:1014af42efd9	84
simon	0:1014af42efd9	85	/* S->pState points to buffer which contains previous frame (numTaps - 1) samples */
simon	0:1014af42efd9	86	/* pStateCurnt points to the location where the new input data should be written */
simon	0:1014af42efd9	87	pStateCurnt = &(S->pState[(numTaps - 1u)]);
simon	0:1014af42efd9	88
simon	0:1014af42efd9	89	/* Initializing blkCnt with blockSize */
simon	0:1014af42efd9	90	blkCnt = blockSize;
simon	0:1014af42efd9	91
simon	0:1014af42efd9	92	while(blkCnt > 0u)
simon	0:1014af42efd9	93	{
simon	0:1014af42efd9	94	/* Copy the new input sample into the state buffer */
simon	0:1014af42efd9	95	pStateCurnt++ = pSrc++;
simon	0:1014af42efd9	96
simon	0:1014af42efd9	97	/* Initialize state pointer */
simon	0:1014af42efd9	98	px = pState;
simon	0:1014af42efd9	99
simon	0:1014af42efd9	100	/* Initialize coefficient pointer */
simon	0:1014af42efd9	101	pb = pCoeffs;
simon	0:1014af42efd9	102
simon	0:1014af42efd9	103	/* Set the accumulator to zero */
simon	0:1014af42efd9	104	acc = 0;
simon	0:1014af42efd9	105
simon	0:1014af42efd9	106	/* Loop unrolling. Process 4 taps at a time. */
simon	0:1014af42efd9	107	tapCnt = numTaps >> 2;
simon	0:1014af42efd9	108
simon	0:1014af42efd9	109	while(tapCnt > 0u)
simon	0:1014af42efd9	110	{
simon	0:1014af42efd9	111	/* Perform the multiply-accumulate */
simon	0:1014af42efd9	112	/* acc += b[N] * x[n-N] */
simon	0:1014af42efd9	113	acc += ((q63_t) (px++)) (*pb++);
simon	0:1014af42efd9	114
simon	0:1014af42efd9	115	/* acc += b[N-1] * x[n-N-1] */
simon	0:1014af42efd9	116	acc += ((q63_t) (px++)) (*pb++);
simon	0:1014af42efd9	117
simon	0:1014af42efd9	118	/* acc += b[N-2] * x[n-N-2] */
simon	0:1014af42efd9	119	acc += ((q63_t) (px++)) (*pb++);
simon	0:1014af42efd9	120
simon	0:1014af42efd9	121	/* acc += b[N-3] * x[n-N-3] */
simon	0:1014af42efd9	122	acc += ((q63_t) (px++)) (*pb++);
simon	0:1014af42efd9	123
simon	0:1014af42efd9	124	/* Decrement the loop counter */
simon	0:1014af42efd9	125	tapCnt--;
simon	0:1014af42efd9	126	}
simon	0:1014af42efd9	127
simon	0:1014af42efd9	128	/* If the filter length is not a multiple of 4, compute the remaining filter taps */
simon	0:1014af42efd9	129	tapCnt = numTaps % 0x4u;
simon	0:1014af42efd9	130
simon	0:1014af42efd9	131	while(tapCnt > 0u)
simon	0:1014af42efd9	132	{
simon	0:1014af42efd9	133	/* Perform the multiply-accumulate */
simon	0:1014af42efd9	134	acc += ((q63_t) (px++)) (*pb++);
simon	0:1014af42efd9	135
simon	0:1014af42efd9	136	/* Decrement the loop counter */
simon	0:1014af42efd9	137	tapCnt--;
simon	0:1014af42efd9	138	}
simon	0:1014af42efd9	139
simon	0:1014af42efd9	140	/* Converting the result to 1.31 format */
simon	0:1014af42efd9	141	/* Store the result from accumulator into the destination buffer. */
simon	0:1014af42efd9	142	acc = (q31_t) (acc >> shift);
simon	0:1014af42efd9	143
simon	0:1014af42efd9	144	*pOut++ = (q31_t) acc;
simon	0:1014af42efd9	145
simon	0:1014af42efd9	146	/* Compute and store error */
simon	0:1014af42efd9	147	e = *pRef++ - (q31_t) acc;
simon	0:1014af42efd9	148
simon	0:1014af42efd9	149	*pErr++ = (q31_t) e;
simon	0:1014af42efd9	150
simon	0:1014af42efd9	151	/* Compute alpha i.e. intermediate constant for taps update */
simon	0:1014af42efd9	152	alpha = (q31_t) (((q63_t) e * mu) >> 31);
simon	0:1014af42efd9	153
simon	0:1014af42efd9	154	/* Initialize state pointer */
simon	0:1014af42efd9	155	/* Advance state pointer by 1 for the next sample */
simon	0:1014af42efd9	156	px = pState++;
simon	0:1014af42efd9	157
simon	0:1014af42efd9	158	/* Initialize coefficient pointer */
simon	0:1014af42efd9	159	pb = pCoeffs;
simon	0:1014af42efd9	160
simon	0:1014af42efd9	161	/* Loop unrolling. Process 4 taps at a time. */
simon	0:1014af42efd9	162	tapCnt = numTaps >> 2;
simon	0:1014af42efd9	163
simon	0:1014af42efd9	164	/* Update filter coefficients */
simon	0:1014af42efd9	165	while(tapCnt > 0u)
simon	0:1014af42efd9	166	{
simon	0:1014af42efd9	167	/* coef is in 2.30 format */
simon	0:1014af42efd9	168	coef = (q31_t) (((q63_t) alpha * (*px++)) >> (32));
simon	0:1014af42efd9	169	/* get coef in 1.31 format by left shifting */
simon	0:1014af42efd9	170	pb = clip_q63_to_q31((q63_t) pb + (coef << 1u));
simon	0:1014af42efd9	171	/* update coefficient buffer to next coefficient */
simon	0:1014af42efd9	172	pb++;
simon	0:1014af42efd9	173
simon	0:1014af42efd9	174	coef = (q31_t) (((q63_t) alpha * (*px++)) >> (32));
simon	0:1014af42efd9	175	pb = clip_q63_to_q31((q63_t) pb + (coef << 1u));
simon	0:1014af42efd9	176	pb++;
simon	0:1014af42efd9	177
simon	0:1014af42efd9	178	coef = (q31_t) (((q63_t) alpha * (*px++)) >> (32));
simon	0:1014af42efd9	179	pb = clip_q63_to_q31((q63_t) pb + (coef << 1u));
simon	0:1014af42efd9	180	pb++;
simon	0:1014af42efd9	181
simon	0:1014af42efd9	182	coef = (q31_t) (((q63_t) alpha * (*px++)) >> (32));
simon	0:1014af42efd9	183	pb = clip_q63_to_q31((q63_t) pb + (coef << 1u));
simon	0:1014af42efd9	184	pb++;
simon	0:1014af42efd9	185
simon	0:1014af42efd9	186	/* Decrement the loop counter */
simon	0:1014af42efd9	187	tapCnt--;
simon	0:1014af42efd9	188	}
simon	0:1014af42efd9	189
simon	0:1014af42efd9	190	/* If the filter length is not a multiple of 4, compute the remaining filter taps */
simon	0:1014af42efd9	191	tapCnt = numTaps % 0x4u;
simon	0:1014af42efd9	192
simon	0:1014af42efd9	193	while(tapCnt > 0u)
simon	0:1014af42efd9	194	{
simon	0:1014af42efd9	195	/* Perform the multiply-accumulate */
simon	0:1014af42efd9	196	coef = (q31_t) (((q63_t) alpha * (*px++)) >> (32));
simon	0:1014af42efd9	197	pb = clip_q63_to_q31((q63_t) pb + (coef << 1u));
simon	0:1014af42efd9	198	pb++;
simon	0:1014af42efd9	199
simon	0:1014af42efd9	200	/* Decrement the loop counter */
simon	0:1014af42efd9	201	tapCnt--;
simon	0:1014af42efd9	202	}
simon	0:1014af42efd9	203
simon	0:1014af42efd9	204	/* Decrement the loop counter */
simon	0:1014af42efd9	205	blkCnt--;
simon	0:1014af42efd9	206	}
simon	0:1014af42efd9	207
simon	0:1014af42efd9	208	/* Processing is complete. Now copy the last numTaps - 1 samples to the
simon	0:1014af42efd9	209	satrt of the state buffer. This prepares the state buffer for the
simon	0:1014af42efd9	210	next function call. */
simon	0:1014af42efd9	211
simon	0:1014af42efd9	212	/* Points to the start of the pState buffer */
simon	0:1014af42efd9	213	pStateCurnt = S->pState;
simon	0:1014af42efd9	214
simon	0:1014af42efd9	215	/* Loop unrolling for (numTaps - 1u) samples copy */
simon	0:1014af42efd9	216	tapCnt = (numTaps - 1u) >> 2u;
simon	0:1014af42efd9	217
simon	0:1014af42efd9	218	/* copy data */
simon	0:1014af42efd9	219	while(tapCnt > 0u)
simon	0:1014af42efd9	220	{
simon	0:1014af42efd9	221	pStateCurnt++ = pState++;
simon	0:1014af42efd9	222	pStateCurnt++ = pState++;
simon	0:1014af42efd9	223	pStateCurnt++ = pState++;
simon	0:1014af42efd9	224	pStateCurnt++ = pState++;
simon	0:1014af42efd9	225
simon	0:1014af42efd9	226	/* Decrement the loop counter */
simon	0:1014af42efd9	227	tapCnt--;
simon	0:1014af42efd9	228	}
simon	0:1014af42efd9	229
simon	0:1014af42efd9	230	/* Calculate remaining number of copies */
simon	0:1014af42efd9	231	tapCnt = (numTaps - 1u) % 0x4u;
simon	0:1014af42efd9	232
simon	0:1014af42efd9	233	/* Copy the remaining q31_t data */
simon	0:1014af42efd9	234	while(tapCnt > 0u)
simon	0:1014af42efd9	235	{
simon	0:1014af42efd9	236	pStateCurnt++ = pState++;
simon	0:1014af42efd9	237
simon	0:1014af42efd9	238	/* Decrement the loop counter */
simon	0:1014af42efd9	239	tapCnt--;
simon	0:1014af42efd9	240	}
simon	0:1014af42efd9	241
simon	0:1014af42efd9	242	}
simon	0:1014af42efd9	243
simon	0:1014af42efd9	244	/**
simon	0:1014af42efd9	245	* @} end of LMS group
simon	0:1014af42efd9	246	*/

Repository toolbox

Export to desktop IDE

Repository details

Type:	Library
Created:	10 Mar 2011
Imports:	907
Forks:	1
Commits:	3
Dependents:	5
Dependencies:	0
Followers:	35

src/Cortex-M4-M3/FilteringFunctions/arm_lms_q31.c@0:1014af42efd9, 2011-03-10 (annotated)

Who changed what in which revision?

Repository toolbox

Repository details

Important Information for this Arm website

Access Warning