An efficient analytical method for dimensioning of CDMA cellular networks serving streaming calls

We propose an efficient analytical method for dimensioning of the downlink radio part of CDMA networks serving real-time calls. The proposed method is based on some admission condition called average feasibility condition (AFC). The advantage of using this condition is twofold: It has the well-known multi-Erlang form making the corresponding call blocking probabilities easy to evaluate, using e.g. Kaufman-Roberts algorithm. Moreover, it approximates the necessary and sufficient condition of the feasibility of power allocation (NSFC) that gives intrinsic, ultimate limitation of the network performance. More precisely, our AFC is some modification of the distributed sufficient condition of the power allocation proposed in [1] which, in its original form, is too conservative yielding a loss of capacity compared to NSFC of about 25% (for the voice traffic). The modification consists of replacing the other-base-station maximal power limitation by an average emitted power approximation evaluated in some simple yet pertinent mean network model. We analytically evaluate the call blocking probabilities in the network model with Poisson arrival stream controlled by AFC. Moreover, in order to validate the pertinence of the proposed approach in cellular network dimensioning process, we compare the obtained blocking probabilities to those estimated from simulations of the model running NSFC (in the regimes of interest for the dimensioning process). This comparison shows that AFC yields a gap of capacity of about 5% (for the voice traffic).