Traffic models and associated parameters in GSM/(E)GPRS networks
Georgeta Budura, Cornel Balint, Adrian Budura, Eugen Mârza · 2009
Abstract:- GSM/(E)GPRS networks support a mixture of traffic consisting of voice and data. A key concept of dimensioning radio resources in such networks is represented by sharing resources between different users and different services. In this paper we address the problem of voice and data traffic models according to different resources allocations strategies: CP (Complete Partitioning) and PP ( Partial Partitioning) and define specific parameters in each case. The proposed traffic model depend in each case on the resources allocation strategy: for voice traffic it is based on both cases, CP and PP, on Erlang law meanwhile for data traffic we build different models depending the scheme used for resources allocation. For CP scheme we propose two models: the model based on Erlang law and those based on modified Engset law. For PP scheme we also proposed two models: one is based on the bi-dimensional Markov chain and the other on the modified Engset law. Both consider voice-data interaction according to the PP scheme. In all cases we define specific performance parameters. We also implement the studied models and evaluate the proposed performance parameters. Finally we compare the results in order to find the adequate model to use for dimensioning purposes. Key-Words:- GSM/(E)GPRS, blocking and preemption probability, Erlang-B law, Engset law, average