How can I make an adaptive thresholding to a signal in order to maintain a constant SNR?

Hi all!

I'm struggling to create an algorithm that gives me a constant SNR from a set of 720x1 vector data. Is there any way I can establish a constant SNR throughout several signals?

The signal is a frequency-averaged Radar Cross Section (RCS) as function of the azimuth angle phi (see images below). The signal of interest is the one plotted with dBm^2, the other isn't the focus of this question. Later, I'm planning to make a simple hypothesis testing (NP) using non-parametric Gaussian KDE. I already have the PDFs but when simple hypothesis is applied, I get a very small confidence interval due to wrong signal thresholding. </matlabcentral/answers/uploaded_files/59258/9060.bmp>

Is there any way of doing this? Any paper, statistical method or matlab function that might give me such output?

RCS=RCS_values; (720x1)

phi=-180:0.5:180; (720x1)

If there's anything missing, please ask!

Thanks in advance!