Impact of mean user speed on blocking and cuts of streaming traffic in cellular networks

This paper deals with the performance evaluation of wireless cellular networks serving streaming traffic. In contrast to data traffic, streaming users require predefined transmission rates, which can be maintained at the price of blocking of some arrivals or cutting some existing connections when a network congestion occurs. The fractions of blocked and cut transmissions in the long run of the system, called respectively, blocking and cut probabilities, are the main performance metrics of such networks. In this paper we evaluate the impact of the intra-and extra-cell user mobility on these metrics. Specifically, we assume a spatio-temporal Poisson arrival process of streaming calls, independent Markovian mobility and exponential duration of each call. The dynamics of this free (offered) process is modified each time a congestion-generating transition occurs. We consider two possible modifications, which lead to two different loss models. We study both of them, and in particular, we propose some explicit approximations for the blocking and cut probabilities, which take into account the mean user mobility speed. We illustrate our approach studying UMTS release 99 building upon the scalable admission control schemes developed in [1], [2], which allow for an exact representation of the geometry of interference in the network.