Targeted Rays at Object

[rudranil2702]

Dear Sionna Team and the users,

I have been doing some Sionna-based simulations and am wondering if it is possible to target a ray to a particular object in the environment. For example, in the attached image, none of the rays hit the gray box, but I want at least one of the rays to hit it.
I have tried increasing the number of ray samples, but this does not solve the problem. When increasing the maximum depth, some rays do hit the target but are not strong enough to be resolved. Please let me know if someone has an experience with this. Thank you in advance and have a nice day! 😄

[merlinND]

Hello @rudranil2702,

Is it possible that there is no valid chain of specular reflections that can connect your transmitter to your receiver via the gray box (at low max depth) ?

To experiment, it could be helpful to open your scene in the new Sionna RT GUI and move the receiver & transmitters around. I imagine that you will find that certain positions do have valid connections.
https://github.com/NVlabs/sionna-rt-gui

[rudranil2702]

Dear @merlinND ,

Thanks a lot for the response. This of course looks like a viable way to find out the optimal placements of the radio devices. 😄
However, I am doing the simulations in an environment and placing the TX and RX and tracking the paths is a challenge due to the size (attached image). I also tried it in the Sionna "box" environment but getting same results. Is there a way to control the sizes of the radio devices? In the outdoor environments, they look fine though.
If not the GUI, maybe through the YAML files? Please let me know.

[merlinND]

Hello @rudranil2702,

Ah, that's unfortunate, we did most of the testing on large scenes so it seems that some of the radii are not adapting well to smaller scenes.

There is no exposed configuration option for paths and radio device visualization size, but you can tweak them easily in the code:

https://github.com/NVlabs/sionna-rt-gui/blob/6d37a263e99cb248468f28eb1417b64cfd70ad34/src/sionna_rt_gui/sionna_utils.py#L354

https://github.com/NVlabs/sionna-rt-gui/blob/6d37a263e99cb248468f28eb1417b64cfd70ad34/src/sionna_rt_gui/sionna_utils.py#L103

By the way, if you have cloned the GUI repository, you can pass --watch to the launch script. It should then automatically detect code changes and reload.

[rudranil2702]

Dear @merlinND,

Thank you very much for your answers. This is indeed quite helpful. I will try it out. 😄

[Vinu2111]

That’s an interesting case, I’ve also noticed that ray interactions can be quite sensitive to geometry and positioning.

From what I understand, if there’s no valid reflection or propagation path, increasing samples alone won’t guarantee a hit.

Exploring receiver/transmitter positioning as suggested sounds like a good approach curious to see how different placements affect the ray paths.

[rudranil2702]

Dear @Vinu2111 , yes, you can check Merlin's comment. Another way could be to run a loop and find optimal TX and RX placements that hit the object of interest. Or probably use a distributed setup which increases the odds of hitting a particular target. 😅

[Vinu2111]

That makes sense especially the idea of iterating over TX/RX placements to discover valid paths.

I was also wondering if combining that with some form of visualization (to track which regions are being covered or missed) could make this process easier to reason about.

Curious if anyone has tried something like that for analyzing ray coverage in complex scenes.