Does anyone have experience with blind source separation and has code they would share?

I am trying to use BSS for EEG signal processing. I only have 4 channels, but the problem is that the data is so noisy simple filtering and ICA are not working. The data isn't human.

I have downloaded several programs from this site for BSS. One has a GUI which is very good, but I cannot understand many things. I am under real time pressure to finish the analysis for the dissertation, but have only just discovered BSS and am very much struggling. If anyone can answer any of these questions I would be VERY grateful! The code referred to below is from matlab central- BSSG by J. Petkov. I found it on file exchange.

1.I don't understand how to interpret the ISR matrix shown before and after testing results. 2. The output does not seem to include the actual components themselves, although supporting documentation says it should be saved as 'signals'. 3. If I only have 4 channels how many components should I be looking for? I suspect there are many blink and muscle artifacts, and possibly cardiac. 4. My goal is to find ERP. Should I be using the time averaged ERP epochs for all 4 channels (4*359 matrix) as the input data or use individual epochs not averaged? 5. Any suggestions on method? Eg. WASOBI etc? Thanks Kim

[ADITI SINGLA]

Good evevning sir/amdam, I need MATLAB code for mixing of two or more signals and then separating it using any algorithm like independent vector(except independent component analysis).OUPUT SHOULD BE EXACT REPLICA OF ORIGINAL SOURCE SIGNAL OR SOMEWHAT DISTORTED ??THANKS IN ADVANCE.