view libgsmefr/convolve.c @ 604:54f0f1b74c25

libgsmhr1 TFO: require BFI=0 and SID=0 for homing In order for a received frame to be recognized as DHF, we need not only the correct bit pattern, but also BFI=0 and SID=0. The BFI=0 requirement should be obvious, while the SID=0 requirement is needed only for HR codec. With FR and EFR, SID classification comes from the payload bits and no separate check is needed - but in HR we get an out-of-band SID ternary flag. When SID=1, no payload bits are used at all; when SID=2, we use only the first 33 bits of the payload. Therefore, it is proper to conditionalize DHF acceptance on SID=0. We already implemented this logic in the just finished full decoder; now bring TFO code into agreement.
author Mychaela Falconia <falcon@freecalypso.org>
date Thu, 04 Dec 2025 19:40:35 +0000
parents 8de2f0f9bd78
children
line wrap: on
line source

/*************************************************************************
 *
 *  FUNCTION:   Convolve
 *
 *  PURPOSE:
 *     Perform the convolution between two vectors x[] and h[] and
 *     write the result in the vector y[]. All vectors are of length L
 *     and only the first L samples of the convolution are computed.
 *
 *  DESCRIPTION:
 *     The convolution is given by
 *
 *          y[n] = sum_{i=0}^{n} x[i] h[n-i],        n=0,...,L-1
 *
 *************************************************************************/

#include "gsm_efr.h"
#include "typedef.h"
#include "namespace.h"
#include "basic_op.h"
#include "no_count.h"
#include "sig_proc.h"

void Convolve (
    Word16 x[],        /* (i)     : input vector                           */
    Word16 h[],        /* (i)     : impulse response                       */
    Word16 y[],        /* (o)     : output vector                          */
    Word16 L           /* (i)     : vector size                            */
)
{
    Word16 i, n;
    Word32 s;

    for (n = 0; n < L; n++)
    {
        s = 0;                  move32 (); 
        for (i = 0; i <= n; i++)
        {
            s = L_mac (s, x[i], h[n - i]);
        }
        s = L_shl (s, 3);
        y[n] = extract_h (s);   move16 (); 
    }

    return;
}