Is my LDPC BER curve right?

Hello,

I am trying to simulate a 3/5 LDPC code. I have encoded it using the Systemic approximate Lower Triangular Encoding system and I am using Sum-Product and Min Sum for Decoding. I have modulated the signal using BPSK and I am passing the signal through the AWGN channel. My error rates only work for negative SNR(dB) values. Occasionally, it does work for positive values (but not more than 1). See images.

But mostly it works for negative values only (Like image 6).

Is this right? I don't know why it happens like this. Can anyone help me with this? Please comment if you need further information. I thank all who takes the time to help me with this.

Answer 1

[vidyesh]

Hi Rishi,

From the figures you've provided, it looks like the BER is quite low for SNR values of 0 and 1 dB, falling below . Since logarithmic graphs can't plot a BER of 0, it's possible that the BER value for higher SNR values is zero, which is why it is not showing up on the plot.

To get a more accurate BER reading for higher SNR values, we could try increasing the number of samples in the simulation. Another approach we can try is to run the simulation multiple times and then average the results to smooth out any anomalies that might occur especially at higher values of SNR where BER is probably very low.

Hope this helps.