confused on perfect CSI simulation result about sionna

[jhoydis]

Hi @vanwayplus,

The BER curve you shared looks identical to the one in the notebook. Could you re-explain what is bothering you?

[vanwayplus]

Hi @jhoydis
Yes, as I understand, the BER of the perfect CSI curve should be always 0 since we already know the accurate status.But in the figure it comes in a curve

[gc1905]

Hi @vanwayplus. The CSI used by the receiver is perfect but not estimated - this is true. However, please remember that the input signal is still affected by AWGN with power indicated on x-axis, which leads to BER degradation. There may be also other reasons for BER degradation depending on specific scenario, e.g., the MIMO channel matrix may be nearly singular.

[SebastianCa]

This issue appears to be solved.

As pointed out by @gc1905, the perfect CSI curve still includes Gaussian noise (otherwise, no SNR could be defined).