[Discuss-gnuradio] SNR Estimator

[Lixing Song]

Hi guys,

Currently, I'm using GNUraido with UHD to do some research about SNR and BER. To estimate the SNR, I use the metric of a paper named Robust Frequency and Timing Synchonization for OFDM," IEEE Trans. Communications, vol. 45, no. 12, 1997, which has been implemented in the ofdm_sync_pn.py file. According to the method of this paper, I get the timing metric( M(d) ) from the binary file "ofdm_sync_pn-mf_f.dat", and the peak signal from "ofdm_sync_pn-peaks_b.dat". Then I sampled the timing metric using peak signal at matlab. After that, taking the metric into the formula (-1 + 1/(1-sqrt(M(d))), which is from the paper.

What my problem is the SNR I got from this method is not sensitive to the signal power changing, I mean, I try to reduce the transmission amplitude to get a bad SNR. But the SNR did not change so much, even I can observe that the frame error increasing sharply. I wonder if you can help me to point out my mistakes, or if you have any better SNR estimator( the mpsk does not work for my research, because I need to change the modulation a lot).

Thanks,

Song