n my model p= probability that a certain user begins transmission in the next slot .  I am trying to match p both in simulation and analytical model.

For simulation, I have counted the transmission attempt of each user and divide them by total simulation time.

p= number of transmission lattempts per user./ total simulation time.

If my number of user are 100 I get 100 ps and then i average them average_p=sum(p)/10.

But the values are quiet different although the trend is same that is decreasing with increase in CW or packet  length where CW=2^(w)* CWmin , CWmin is initial Contention window  and w is backoff stage depending on the number of unsuccessful transmission, with each unsuccessful transmission Cw doubles.