The first model that you use is a multinomial model, which assume that the voltage level
in each time unit is independent from other.
The second model that you are concern of is like some time series data that each data
in the series are actually correlated with one or more data in that series.
So you better check the physical properties that the assumption of independence is
valid or not. If thats not proper to make such assumption, then we have to investigate
a more complicated "time series" model.