problem in channel decoder implementation using sionna

[Sapta78342]

I am trying to design a Deep RX, and try to calculate Coded BER (After channel decoder at Rx) and Uncoded BER (before channel decoder at RX) For the uncoded case I am getting a perfect plot Baseline fully known channel perform better than the deep rx

But for coded Ber I am getting different ? Can any one suggest why is this happening ?

[SebastianCa]

Hi @Sapta78342,

I’m not entirely sure I understand your plots. The coding gains seem quite small—what is the block length of the code? Are you using LDPC codes? Also, what’s the difference in the setup between the left and right figures? Have you compared your results to the Neural Receiver Tutorial Notebook?