Signal Processing Stack Exchange is a question and answer site for practitioners of the art and science of signal, image and video processing. It only takes a minute to sign up.
Sign up to join this community
I am sampling a 5mhz blocks at an IF of 40 MHZ. I am sampling at 11MHz per bandpass sampling theory, but when I run the FFT I get all the aliases.
So I am trying to figure out how to constrain the fft interpretation to the appropriate bands.
IE i may be sampling the band from 80-85MHz by mixing it down to 37.5-42.5 MHZ and sampling at 11 MHZ. How do I take the FFT and plot it showing the peaks appropriately?
If there is other frequency content in other Nyquist regions, you will see those alias into your digitized signal also; there's no free lunch here. If you wish to use bandpass sampling, then you need to first apply a bandpass anti-aliasing filter to suppress the unwanted content.
Look at this paper by Emmanuel Candés. It describes exactly how to sample some signal at a low frequency (11MHz) when it's at a higher frequency (40MHz).
A compressed sensing technique is called "iterative hard thresholding." It's represented by $x^{n+1} = H_K(x^n + FFT(y-FFT^{-1}(x^n)) $ where $H_K$ is a nonlinear operator that takes the largest $k$ terms, $y$ your measurements in the time domain, and $x$ your reconstruction in the Fourier domain.
I made a gist available. I was using the discrete wavelet transform, but you should easily be able to replace that with the FFT. This gist reads an image then reconstructs it, and does the associated setup.
This may not be what you're looking for: you may not be able to do the computation on the fly like this. My bet is that you could do these computations when the signal is returned to the "brain," but that will only work if it's not real-time (although it may work anyways).
