Study and Analysis of Cellular Wireless Networks with Multiclass Traffic

In this paper, the traffic characteristics have been studied by collecting traces from a CDMA2000 cellular wireless network that provides services like messaging, video streaming, e-mail, internet, song downloading. The traces record call activities including call initiation time, termination time, originating node identification number, packet size, home station id, foreign station id (when roaming), handoffs. Traffic parameters namely, call inter-arrival times and call holding times were estimated using statistical methods. The results show that call inter-arrival time distribution in this CDMA cellular wireless network is heavy-tailed and can be modeled by gamma as well as weibull distributions and are asymptotically long-range dependent. It is also found that the call holding times are best fitted with lognormal distribution and are not correlated. An analytical model based on our observations for performance measures of a circuit-switched cellular wireless network with multiclass traffic sources is also proposed.