Using smoothing functions without incoorporating NaN/interpolating over NaN

Hi

I have a time series consisting of time and pairs of long and lat positions. Due to errors there are data gaps in my time series but i have filled them out with NaN so I now have a continuous time series for a whole year with positions every third hour. What I want to do is to smooth my data using one of the functions in matlab (currently a moving average), but when I do so and then plot my data there are no data gaps so i Guess that the function interpolate over my data gaps? Is that correct and if, can I make it not interpolate, caus I want the data gaps to still be there... I have attached a picture of my long/lat pairs before and after smoothing and as you can see there are data gaps present before the smoothing.... Any help would be lovely as I am new to matlab....