Problem with the high correlation in noise while receiving in USRP


I am trying to implement Eigenvalue based sensing method using gnuradio
USRP. and while doing the testing, when we try to receive the noise and
the detection we still get the output as 1 (i.e. the ratio of eigevalues
higher than threshold). This may be due to the high correlation in
noise. I
tried with removing the antenna and also tried increasing the gain but
the problem is there. I can someone help me with this problem.

Thanking you,



I’m also tryng to implement a eigenvalue based sensing method and I’m
having a similar problem. How are you chosing the threshold? And what
size is your matrix and whith how many samples are you doing the

The problem could be in the chosing of the threshold if you used the
assyptotic or semi-assymptotic way. But also when you are decimating the
signal, you are also filtering it and getting noise that is no longer
white, thus correlated.

Did you solve the problem? did you try pre whitening the signal?
I am assuming that your trying to use oversampling, what is your
sampling frequency and decimation?

Regards Pedro