A New Statistical Model for Call Holding Time Simulation in the Telephony Networks

Call interval time and call holding time patterns are two important topics in the traffic analysis in telephony networks. In this paper to simulate call holding time, we compare a real case of voice call holding time distribution with statistical models and effort to find the best model which has more similarity to real data. We use exponential, lognormal, Erlang-k, Erlang-jk and hyper Erlang-jk distributions for simulating the call holding time. Results of the simulations shows that the Hyper Erlang-jk has the least error and is more suitable for this purpose. By changing the structure of this model, we present a new model that has less error and more similarity to real data. Finally, we introduce a practical sample of using the proposed model for generating traffic for implementation performance test procedures in signaling type approval platform (STAP) implemented at testing laboratories of Iran Telecom Research Center (ITRC).