# Is there a way to apply blind source separation using ONE mixture?

This question is a follow-up for my previous one: Reduce the "glare" effect of a receiver within a system.

I have broken the problem into pieces and I humbly think that I've got a way to solve it it with Duet BSS. However, there is still a problem: I have got

$$Y(t)=h_1(t)\otimes X(t) + h_2(t)\otimes s(t)$$

Is there a way be it a hardware or a software solution to use ONE mixture to get my two signals of which I only know that $X$ is much stronger than $s$ and that on the spectrogram they don't overlap?

I thought about putting a relay mast between the emitters and the receptor so I can get two rich mixtures. However, there is a risk that I increase, by doing so, the "blinding" effect on the receptor.

The signals are not periodic but could be seen as cyclostationary.

• What is the nature of x(t) and s(t)? Is one a tone interferer, and another a pulsed signal for example? – Tarin Ziyaee Jul 8 '14 at 18:30
• I don't have an info on them, just that they could be considered cyclostationary. Is there a way if one of them is a pulsed signal?(Just curious) – Haydie Jul 8 '14 at 20:57
• Well, on what criteria do you want to separate them by? Just the fact that there are two signals? – Tarin Ziyaee Jul 9 '14 at 14:32
• Thanks for your input, and yes I don't have any special criteria by which I want to separate them...Maybe by taking into consideration that one is stronger than the other, but aparat from that I see nothing else. – Haydie Jul 9 '14 at 14:57
• You have to specify how you want to discriminate between them though, otherwise there is really nothing to go off of. Or you would specify some sort of statistical property that they both have or dont have, or something of that sort. – Tarin Ziyaee Jul 9 '14 at 15:14