AWR1843: question about aoa

Hello,Cesar

I can not open the link you offered.I find the atan2 function that define as follow:

Domain is .

However the code is as follow:

Wz = (float) atan2sp(tempIm, tempRe) * (1.0f/PI_);  //Domain is (-1,1];
if (Wz > 1)
{
    Wz = Wz - 2;
}
else if (Wz < -1)
{
    Wz = Wz + 2;
}
z = range * Wz;

So why will Wz bigger than 1?

Thank you

Bryant

Hi,

The code is as follows

Thank you

Cesar

So first the variables Wx and Wz that we have in the code are normalized phases Wx an Wz shown in formulas. I think we mention that somewhere in the code.

So let’s start for Wz:

Basic proportion:

Wz = 2*pi/lambda * AB

where AB = d*sin(phi)    (from figure in doxygen, where we should replace (lambda/2) with d)

So we have:

Wz = 2*pi/lambda * d *sin(phi)       (eq1)

Coordinate Z relation to spherical:

Z = R * sin(phi)                                   (eq2)

Now replace sin(phi) in eq2 from eq1

Z = R * lambda/(2*d) * (Wz/pi)

So we should add above correction when we calculate Z, not Wz.

Where Wz/pi is calculated as atan2(im,re) . (I remember the formula was:  atan2(im,re)+2Wx and that is why we check if it s greater than 1 we subtract 2. This +2Wx I do not see in the line highlighted in yellow below)

Now let’s look for Wx:

Wx = 2*pi/lambda * CD

Where CD = d * sin(theta)*cos(phi) (from figure in doxygen)

So we have:

Wx = 2*pi/lambda * d * sin(theta)*cos(phi)          (eq3)

From FFT we calculate Wx as:

Wx = (2*pi/N) * Kmax                                            (eq4)

Now replace Wx in eq3 from eq4 :

(2*pi/N) * Kmax = 2*pi/lambda * d * sin(theta)*cos(phi)    (eq5)

The coordinate x relation to spherical:

X = R * sin(theta)*cos(phi)                              (eq6)

Now replace sin(theta)*cos(phi) from eq5 into eq6:

X = R * lambda/(2*d) * 2*Kmax/N

I found some old code in postprocessing.c and place there the corrections (in green):

Please check, I think this should be right. We can have a call next week if you want.

Regards,

Slobodan

    Wx = 2 * (float) sMaxIdx / numAngleBins;

    x = range * (lambda/(2*d)) * Wx;

    if (obj->numVirtualAntElev > 0)

    {

        Wz = atan2(peakAzimIm * peakElevRe - peakAzimRe * peakElevIm,

                   peakAzimRe * peakElevRe + peakAzimIm * peakElevIm)/PI_ + (2 * Wx);

        if (Wz > 1)

        {

              Wz = Wz - 2;

        }

        else if (Wz < -1)

        {

              Wz = Wz + 2;

        }

        z = range * lambda/(2*d)  * Wz;

        /*record wz for debugging/testing*/

        if(objOutIndx < MMW_MAX_ELEV_OBJ_DEBUG)

        {

            obj->detObjElevationAngle[objOutIndx] = Wz;

        }

    }

    else

    {

        z = 0;

    }

Hello,Cesar

Thank you very much for your answer. I understand.

BR

Bryant