Separating background data from expected data

[majormuss]

Hi all,
I was hoping if someone can tell me what's the best way to separate background from expected. I have attached a picture of a representative graph. The data points are not real but the behavior is similar to what I expect to get from my research. The data I expect is supposed to be approximately constant across the graph and the background is more variable. Hence, the background in my picture will probably be the peaks. However, I am not sure what statistical method to use or which would be best suited for my research.

 

[mfb]

You will need more information about the background. Some estimate for the shape, the magnitude or something else.

 

[majormuss]

You will need more information about the background. Some estimate for the shape, the magnitude or something else.

I have made changes to my question. The background are the peaks and the expected data is approximately flat.

 

[mfb]

Which peaks? Looks like random fluctuations to me.

Do you expect a constant signal plus random fluctuations around this signal, centered at the signal or at some other value? Then your best estimate for the signal is the average of all those points, minus the offset if the background has one.
The description is still very vague.