Possible AoD Bug for Multi-Bounce Paths (max_depth=2)

[tlstkddn999]

Hello everyone,

I'm currently using Sionna RT version 1.0.2, and I’ve encountered what seems to be an inconsistency in the computation of the AoD when max_depth is set to 2.

From my understanding, the AoD should always represent the angle from the TX to the first bounce point (BP1) in the global coordinate system (GCS), since that is the direction in which the wave departs the TX.

However, during debugging, I’ve noticed that in some cases when max_depth=2, the stored AoD appears to instead represent the angle from BP1 to BP2 (i.e., the first reflection point to the second), rather than from TX to BP1. This does not occur consistently but happens intermittently for certain paths.

To clarify:

• For a path like TX → BP1 → BP2 → RX, I expect AoD to represent TX → BP1.

• But occasionally, it seems to represent BP1 → BP2, which is likely incorrect.

This issue does not seem to occur when max_depth = 1.

Is this a known issue, or could it be a bug in the AoD assignment logic for multi-bounce paths?

For clearer illustration, the attached image was generated using a MATLAB simulation environment

A correct-case example is also provided for reference.

[faycalaa]

Hi @tlstkddn999,

Could you provide a minimal reproducer for this bug?