Develop a simultaneous model for four or six datasets

What I need to do is: develop a simultaneous Negative Binomial model for 4 time periods and account for the correlation between the time periods and the repetition of the observations. Here are more details about my data and what I want. My data: my data is recorded every 15 minutes. I divided the whole day data (96 observations within the whole day) into 4 time periods to see the effect of time on these observations. My time periods are (Time period I: 2 hours with 8 observations; Time period II: 7 hours with 28 observations; Time period III: 2 hours with 8 observations; and Time period IV: 13 hours with 52 observations). I repeat this procedure 68 times (i.e. 68 days). The attached file shows the time periods and the number of observations with the hours as its explained above. The dependent variable is integer value greater then or equal to zero. The independent variables are a combination of ( integer values, binary values, real positive values), some of these independent variables are repeated in each record per day. So, I want to account for that in my model (account to the repetition in my data) . The other thing that I want to account for it in the model is the correlation between the time periods. Since time period II comes after time period I, then there is a correlation between time period I and II and so on for other time periods. If you know how to write a SAS or R code for this and want more detail or explaination, Please do not hesitate to contact me. Thanks