FIRLMSOne_Direct

Filters a single sample through a FIR LMS filter.

Syntax

IppStatus ippsFIRLMSOne_Direct_32f(Ipp32f src, Ipp32f refVal, Ipp32f* pDstVal, Ipp32f* pTapsInv, int tapsLen, float mu, Ipp32f* pDlyLine, int* pDlyLineIndex);

IppStatus ippsFIRLMSOne_Direct32f_16s(Ipp16s src, Ipp16s refVal, Ipp16s* pDstVal, Ipp32f* pTapsInv, int tapsLen, float mu, Ipp16s* pDlyLine, int* pDlyLineIndex);

IppStatus ippsFIRLMSOne_DirectQ15_16s(Ipp16s src, Ipp16s refVal, Ipp16s* pDstVal, Ipp32s* pTapsInv, int tapsLen, int muQ15, Ipp16s* pDlyLine, int* pDlyLineIndex);

Parameters

src	Input sample to be filtered.
pDstVal	Pointer to the output sample.
refVal	Reference signal sample.
pTapsInv	Pointer to the array containing the FIR filter taps to be adapted. The tap values are stored in the array in the inverse order.
tapsLen	Number of elements in the array containing the tap values.
pDlyLine	Pointer to the array holding the delay line values.
pDlyLineIndex	Pointer to the current index of the delay line.
mu	Adaptation step.
muQ15	Integer version adaptation step.

Description

The function ippsFIRLMSOne_Direct is declared in the ipps.h file. This function performs directly a single iteration of FIR filter taps adaptation. The tapsLen-length array pTapsInv contains the FIR filter taps in the inverse order. The (2*tapsLen)-length array pDlyLine specifies the delay line values. The pDlyLineIndex array specifies the current index of the delay line. The output signal is stored in pDstVal.

Note

The adaptation error value can be computed as follows: err[n] = refVal[n] - *pDstVal.

The function ippsFIRLMSOne_Direct performs a single iteration of FIR filter taps adaptation with the mu step value. The taps are floating-point numbers.

Set the taps to zero or to values close to calculated to speed up the process.

The function ippsLMSOne_Direct is to be called within cycle with the number of iterations equal to the number of input samples. You can decide what data is to be saved as a result of adaptation: either the filtered output signal or the adaptation error.

The function ippsFIRLMSOne_DirectQ15 performs a single iteration of FIR filter taps adaptation with the muQ15 step value. The taps are integer numbers. The adaptation step muQ15 can be computed as follows:

muQ15 = (int)(mu *(1<<15)+0.5f)

The example belowExample 6-13 illustrates the use of the function ippsFIRLMSOne_Direct to adapt the FIR filter taps. After adaptation the taps are close to 1.0.

Return Values

ippStsNoErr	Indicates no error.
ippStsNullPtrErr	Indicates an error when the pointers to data arrays are NULL.
ippStsSizeErr	Indicates an error when tapsLen is less or equal to 0.

Using the Function ippsLMSOne_Direct

IppStatus firlmsone(void) {

#define LEN 200

      IppStatus st = ippStsNoErr;

      Ipp32f taps = 0, dly[2] = {0};

      Ipp32f x[LEN], y[LEN], mu = 0.05f;

      int i, indx = 0;

      /// make a const signal of amplitude 1 and noise it

      for( i=0; i<LEN; ++i ) x[i] = 0.4f * rand()/RAND_MAX + 0.8f;

      for( i=0; i<LEN-1 && ippStsNoErr==st; ++i )

      st=ippsFIRLMSOne_Direct_32f( x[i], x[i+1], y+1+i, &taps, 1, mu,

          dly, &indx );

      printf_32f(“FIRLMSOne tap adapted =”, &taps, 1, st );

      return st;

Output:

    FIRLMSOne tap adapted = 0.993872