arm_iir_lattice_f32.c

00001 /* ----------------------------------------------------------------------  
00002 * Copyright (C) 2010 ARM Limited. All rights reserved.  
00003 *  
00004 * $Date:        29. November 2010  
00005 * $Revision:    V1.0.3  
00006 *  
00007 * Project:      CMSIS DSP Library  
00008 * Title:        arm_iir_lattice_f32.c  
00009 *  
00010 * Description:  Floating-point IIR Lattice filter processing function.  
00011 *  
00012 * Target Processor: Cortex-M4/Cortex-M3
00013 *  
00014 * Version 1.0.3 2010/11/29 
00015 *    Re-organized the CMSIS folders and updated documentation.  
00016 *   
00017 * Version 1.0.2 2010/11/11  
00018 *    Documentation updated.   
00019 *  
00020 * Version 1.0.1 2010/10/05   
00021 *    Production release and review comments incorporated.  
00022 *  
00023 * Version 1.0.0 2010/09/20   
00024 *    Production release and review comments incorporated  
00025 *  
00026 * Version 0.0.7  2010/06/10   
00027 *    Misra-C changes done  
00028 * -------------------------------------------------------------------- */ 
00029  
00030 #include "arm_math.h" 
00031  
00032 /**  
00033  * @ingroup groupFilters  
00034  */ 
00035  
00036 /**  
00037  * @defgroup IIR_Lattice Infinite Impulse Response (IIR) Lattice Filters  
00038  *  
00039  * This set of functions implements lattice filters  
00040  * for Q15, Q31 and floating-point data types.  Lattice filters are used in a   
00041  * variety of adaptive filter applications.  The filter structure has feedforward and  
00042  * feedback components and the net impulse response is infinite length.  
00043  * The functions operate on blocks  
00044  * of input and output data and each call to the function processes  
00045  * <code>blockSize</code> samples through the filter.  <code>pSrc</code> and  
00046  * <code>pDst</code> point to input and output arrays containing <code>blockSize</code> values.  
00047   
00048  * \par Algorithm:  
00049  * \image html IIRLattice.gif "Infinite Impulse Response Lattice filter"  
00050  * <pre>  
00051  *    fN(n)   =  x(n)  
00052  *    fm-1(n) = fm(n) - km * gm-1(n-1)   for m = N, N-1, ...1  
00053  *    gm(n)   = km * fm-1(n) + gm-1(n-1) for m = N, N-1, ...1  
00054  *    y(n)    = vN * gN(n) + vN-1 * gN-1(n) + ...+ v0 * g0(n)  
00055  * </pre>  
00056  * \par  
00057  * <code>pkCoeffs</code> points to array of reflection coefficients of size <code>numStages</code>.   
00058  * Reflection coefficients are stored in time-reversed order.  
00059  * \par  
00060  * <pre>  
00061  *    {kN, kN-1, ....k1}  
00062  * </pre>  
00063  * <code>pvCoeffs</code> points to the array of ladder coefficients of size <code>(numStages+1)</code>.   
00064  * Ladder coefficients are stored in time-reversed order.  
00065  * \par  
00066  * <pre>  
00067  *    {vN, vN-1, ...v0}  
00068  * </pre>  
00069  * <code>pState</code> points to a state array of size <code>numStages + blockSize</code>.  
00070  * The state variables shown in the figure above (the g values) are stored in the <code>pState</code> array.  
00071  * The state variables are updated after each block of data is processed; the coefficients are untouched.  
00072  * \par Instance Structure  
00073  * The coefficients and state variables for a filter are stored together in an instance data structure.  
00074  * A separate instance structure must be defined for each filter.  
00075  * Coefficient arrays may be shared among several instances while state variable arrays cannot be shared.  
00076  * There are separate instance structure declarations for each of the 3 supported data types.  
00077   *  
00078  * \par Initialization Functions  
00079  * There is also an associated initialization function for each data type.  
00080  * The initialization function performs the following operations:  
00081  * - Sets the values of the internal structure fields.  
00082  * - Zeros out the values in the state buffer.  
00083  *  
00084  * \par  
00085  * Use of the initialization function is optional.  
00086  * However, if the initialization function is used, then the instance structure cannot be placed into a const data section.  
00087  * To place an instance structure into a const data section, the instance structure must be manually initialized.  
00088  * Set the values in the state buffer to zeros and then manually initialize the instance structure as follows:  
00089  * <pre>  
00090  *arm_iir_lattice_instance_f32 S = {numStages, pState, pkCoeffs, pvCoeffs};  
00091  *arm_iir_lattice_instance_q31 S = {numStages, pState, pkCoeffs, pvCoeffs};  
00092  *arm_iir_lattice_instance_q15 S = {numStages, pState, pkCoeffs, pvCoeffs};  
00093  * </pre>  
00094  * \par  
00095  * where <code>numStages</code> is the number of stages in the filter; <code>pState</code> points to the state buffer array;  
00096  * <code>pkCoeffs</code> points to array of the reflection coefficients; <code>pvCoeffs</code> points to the array of ladder coefficients.  
00097  * \par Fixed-Point Behavior  
00098  * Care must be taken when using the fixed-point versions of the IIR lattice filter functions.  
00099  * In particular, the overflow and saturation behavior of the accumulator used in each function must be considered.  
00100  * Refer to the function specific documentation below for usage guidelines.  
00101  */ 
00102  
00103 /**  
00104  * @addtogroup IIR_Lattice  
00105  * @{  
00106  */ 
00107  
00108 /**  
00109  * @brief Processing function for the floating-point IIR lattice filter.  
00110  * @param[in] *S points to an instance of the floating-point IIR lattice structure.  
00111  * @param[in] *pSrc points to the block of input data.  
00112  * @param[out] *pDst points to the block of output data.  
00113  * @param[in] blockSize number of samples to process.  
00114  * @return none.  
00115  */ 
00116  
00117 void arm_iir_lattice_f32( 
00118   const arm_iir_lattice_instance_f32 * S, 
00119   float32_t * pSrc, 
00120   float32_t * pDst, 
00121   uint32_t blockSize) 
00122 { 
00123   float32_t fcurr, fnext, gcurr, gnext;          /* Temporary variables for lattice stages */ 
00124   float32_t acc;                                 /* Accumlator */ 
00125   uint32_t blkCnt, tapCnt;                       /* temporary variables for counts */ 
00126   float32_t *px1, *px2, *pk, *pv;                /* temporary pointers for state and coef */ 
00127   uint32_t numStages = S->numStages;             /* number of stages */ 
00128   float32_t *pState;                             /* State pointer */ 
00129   float32_t *pStateCurnt;                        /* State current pointer */ 
00130  
00131   gcurr = 0.0f; 
00132   blkCnt = blockSize; 
00133  
00134   pState = &S->pState[0]; 
00135  
00136   /* Sample processing */ 
00137   while(blkCnt > 0u) 
00138   { 
00139     /* Read Sample from input buffer */ 
00140     /* fN(n) = x(n) */ 
00141     fcurr = *pSrc++; 
00142  
00143     /* Initialize state read pointer */ 
00144     px1 = pState; 
00145     /* Initialize state write pointer */ 
00146     px2 = pState; 
00147     /* Set accumulator to zero */ 
00148     acc = 0.0f; 
00149     /* Initialize Ladder coeff pointer */ 
00150     pv = &S->pvCoeffs[0]; 
00151     /* Initialize Reflection coeff pointer */ 
00152     pk = &S->pkCoeffs[0]; 
00153  
00154  
00155     /* Process sample for first tap */ 
00156     gcurr = *px1++; 
00157     /* fN-1(n) = fN(n) - kN * gN-1(n-1) */ 
00158     fnext = fcurr - ((*pk) * gcurr); 
00159     /* gN(n) = kN * fN-1(n) + gN-1(n-1) */ 
00160     gnext = (fnext * (*pk++)) + gcurr; 
00161     /* write gN(n) into state for next sample processing */ 
00162     *px2++ = gnext; 
00163     /* y(n) += gN(n) * vN  */ 
00164     acc += (gnext * (*pv++)); 
00165  
00166     /* Update f values for next coefficient processing */ 
00167     fcurr = fnext; 
00168  
00169     /* Loop unrolling.  Process 4 taps at a time. */ 
00170     tapCnt = (numStages - 1u) >> 2; 
00171  
00172     while(tapCnt > 0u) 
00173     { 
00174       /* Process sample for 2nd, 6th ...taps */ 
00175       /* Read gN-2(n-1) from state buffer */ 
00176       gcurr = *px1++; 
00177       /* Process sample for 2nd, 6th .. taps */ 
00178       /* fN-2(n) = fN-1(n) - kN-1 * gN-2(n-1) */ 
00179       fnext = fcurr - ((*pk) * gcurr); 
00180       /* gN-1(n) = kN-1 * fN-2(n) + gN-2(n-1) */ 
00181       gnext = (fnext * (*pk++)) + gcurr; 
00182       /* y(n) += gN-1(n) * vN-1  */ 
00183       /* process for gN-5(n) * vN-5, gN-9(n) * vN-9 ... */ 
00184       acc += (gnext * (*pv++)); 
00185       /* write gN-1(n) into state for next sample processing */ 
00186       *px2++ = gnext; 
00187  
00188  
00189       /* Process sample for 3nd, 7th ...taps */ 
00190       /* Read gN-3(n-1) from state buffer */ 
00191       gcurr = *px1++; 
00192       /* Process sample for 3rd, 7th .. taps */ 
00193       /* fN-3(n) = fN-2(n) - kN-2 * gN-3(n-1) */ 
00194       fcurr = fnext - ((*pk) * gcurr); 
00195       /* gN-2(n) = kN-2 * fN-3(n) + gN-3(n-1) */ 
00196       gnext = (fcurr * (*pk++)) + gcurr; 
00197       /* y(n) += gN-2(n) * vN-2  */ 
00198       /* process for gN-6(n) * vN-6, gN-10(n) * vN-10 ... */ 
00199       acc += (gnext * (*pv++)); 
00200       /* write gN-2(n) into state for next sample processing */ 
00201       *px2++ = gnext; 
00202  
00203  
00204       /* Process sample for 4th, 8th ...taps */ 
00205       /* Read gN-4(n-1) from state buffer */ 
00206       gcurr = *px1++; 
00207       /* Process sample for 4th, 8th .. taps */ 
00208       /* fN-4(n) = fN-3(n) - kN-3 * gN-4(n-1) */ 
00209       fnext = fcurr - ((*pk) * gcurr); 
00210       /* gN-3(n) = kN-3 * fN-4(n) + gN-4(n-1) */ 
00211       gnext = (fnext * (*pk++)) + gcurr; 
00212       /* y(n) += gN-3(n) * vN-3  */ 
00213       /* process for gN-7(n) * vN-7, gN-11(n) * vN-11 ... */ 
00214       acc += (gnext * (*pv++)); 
00215       /* write gN-3(n) into state for next sample processing */ 
00216       *px2++ = gnext; 
00217  
00218  
00219       /* Process sample for 5th, 9th ...taps */ 
00220       /* Read gN-5(n-1) from state buffer */ 
00221       gcurr = *px1++; 
00222       /* Process sample for 5th, 9th .. taps */ 
00223       /* fN-5(n) = fN-4(n) - kN-4 * gN-1(n-1) */ 
00224       fcurr = fnext - ((*pk) * gcurr); 
00225       /* gN-4(n) = kN-4 * fN-5(n) + gN-5(n-1) */ 
00226       gnext = (fcurr * (*pk++)) + gcurr; 
00227       /* y(n) += gN-4(n) * vN-4  */ 
00228       /* process for gN-8(n) * vN-8, gN-12(n) * vN-12 ... */ 
00229       acc += (gnext * (*pv++)); 
00230       /* write gN-4(n) into state for next sample processing */ 
00231       *px2++ = gnext; 
00232  
00233       tapCnt--; 
00234  
00235     } 
00236  
00237     fnext = fcurr; 
00238  
00239     /* If the filter length is not a multiple of 4, compute the remaining filter taps */ 
00240     tapCnt = (numStages - 1u) % 0x4u; 
00241  
00242     while(tapCnt > 0u) 
00243     { 
00244       gcurr = *px1++; 
00245       /* Process sample for last taps */ 
00246       fnext = fcurr - ((*pk) * gcurr); 
00247       gnext = (fnext * (*pk++)) + gcurr; 
00248       /* Output samples for last taps */ 
00249       acc += (gnext * (*pv++)); 
00250       *px2++ = gnext; 
00251       fcurr = fnext; 
00252  
00253       tapCnt--; 
00254  
00255     } 
00256  
00257  
00258     /* y(n) += g0(n) * v0 */ 
00259     acc += (fnext * (*pv)); 
00260  
00261     *px2++ = fnext; 
00262  
00263     /* write out into pDst */ 
00264     *pDst++ = acc; 
00265  
00266     /* Advance the state pointer by 4 to process the next group of 4 samples */ 
00267     pState = pState + 1u; 
00268     blkCnt--; 
00269  
00270   } 
00271  
00272   /* Processing is complete. Now copy last S->numStages samples to start of the buffer  
00273      for the preperation of next frame process */ 
00274  
00275   /* Points to the start of the state buffer */ 
00276   pStateCurnt = &S->pState[0]; 
00277   pState = &S->pState[blockSize]; 
00278  
00279   tapCnt = numStages >> 2u; 
00280  
00281   /* copy data */ 
00282   while(tapCnt > 0u) 
00283   { 
00284     *pStateCurnt++ = *pState++; 
00285     *pStateCurnt++ = *pState++; 
00286     *pStateCurnt++ = *pState++; 
00287     *pStateCurnt++ = *pState++; 
00288  
00289     /* Decrement the loop counter */ 
00290     tapCnt--; 
00291  
00292   } 
00293  
00294   /* Calculate remaining number of copies */ 
00295   tapCnt = (numStages) % 0x4u; 
00296  
00297   /* Copy the remaining q31_t data */ 
00298   while(tapCnt > 0u) 
00299   { 
00300     *pStateCurnt++ = *pState++; 
00301  
00302     /* Decrement the loop counter */ 
00303     tapCnt--; 
00304   } 
00305  
00306 } 
00307  
00308  
00309  
00310  
00311 /**  
00312  * @} end of IIR_Lattice group  
00313  */