Um, no, you can't. Or, well, least squares may be inappropriate. Or, yes, it is completely possible. Anything in that spectrum of answers is true. (No pun intended.)
The fact is, you need a model before you even consider fitting anything with "least squares". You don't have a model. You have not posed one, in fact, odds are you really probably don't know what that means. So no, you cannot use least squares.
At the other end of the spectrum, once you have chosen a model, once you decide if the noise in your data is meaningful to try to treat using least squares, then yes, of course you can fit it.
So, first, you need to understand how one might develop a model for such a system. Of course, that is wildly beyond what is on-topic for this forum. It has absolutely nothing to do with MATLAB.
Next, the model that you choose often tends to suggest a specific method for how one might fit that model. Is least squares even appropriate? Not always. For example, some noise structures tend to be proportional error, in which case least squares on the log of the data is often a better idea. Normally distributed noise is not always the correct assumption, and that is a fundamental assumption behind least squares estimation. Sometimes maximum likelihood estimation is a better idea.
But until you choose a model for that process, thinking about how you would perform the fit is putting the cart before the horse.
Do you want a better, more specific answer? Not really possible, since until you get more specific about your question, including the model you would pose, I'd argue no more complete answer makes sense.
