Simulating radar multipath reflection?

[mlouielu]

Hi there,

I want ask for a radar simulation question, for a ULA antenna, can sionna do something similar to this: https://www.mathworks.com/help/driving/ug/radar-ghost-multipath.html

I'm not sure if this is related to #844, that I will need to have a custom path resolver and a custom SBR to limit ray shooting FOV?

Thanks for any inputs!

[mlouielu]

I think I will need to customize spawn_raw_from_sources, spawn_ry_to, or to add a mask or antenna gain pattern to the ray gen result?

[faycalaa]

Hi @mlouielu,

To restrict the directions of departures of paths to a specific FOV, you could use another lattice in spawn_ray_from_sources instead of the Fibonacci one.

Note that if your lattice does not ensure that all ray tubes share the same fraction of the unit sphere $$4\pi/N$$, where $$N$$ is the number of samples, you will need to ensure that the solid angle is set accordingly in the field calculation for diffuse reflection to be correctly computed.

A simpler solution could be to use a directive antenna pattern (such as tr38901) to ensure that only paths with the desired directions of departure carry significant energy.

Also, to ensure the receiver observes reflected paths, you could enable diffuse reflection for the scatterers by setting the scattering_coefficient of their radio materials to a value higher than zero.

Hope this helps.

[mlouielu]

Thanks for the reply!

If I tried to use antenna pattern to simulate this, does that mean I will 1) get the same paths from path_solver, 2) but having paths.a scaling with my antenna pattern?

Because after checking the code, it seems that antenna pattern only applied when doing the field calculation?

Thanks!

[faycalaa]

Yes.

The path coefficients paths.a depend on both the transmitter and receiver antenna patterns. However, the path vertices (the intersection points of the paths with the scatterers) do not depend on the antenna patterns.

[mlouielu]

Thank you!

So if I have a vertical polarized patch antenna, with it's E-plane & H-plane antenna gain pattern, the v_pattern should look like this?

def v_pattern(theta, phi):
  c_theta = calc_theta_gain(theta)
  c_phi = calc_phi_gain(phi)
  c = c_theta * c_phi
  return mi.Complex2f(c, 0.0)

And the antenna pattern will be like:

PolarizedAntennaPattern(v_patter=v_patter, polarization="V", polarization_model="tr38901_2")

[mlouielu]

For the customize ray candidate, can I still use fib lattice, but with different warping method (e.g. square_to_uniform_cone(), or customize a square_to_uniform_fov_cone()), and based on this warping method to calculate the solid angle accordingly?