Delay performance of data traffic in a cellular wireless ATM network

Abstract

The delay performance for self-similar data traffic over a prioritized Time Division Multiple Access with Dynamic Reservation (TDMA/DR) MAC protocol in a wireless ATM setting is evaluated using simulation and approximate analysis. An approximate analysis of heavy-tailed ON/OFF data traffic yields very accurate mean delay and delay distribution of the data traffic in the traffic range up to 70% of the total channel capacity. The reasonableness of the approximate analysis is verified by simulation. The aggregate of the heavy-tailed ON/OFF sources, with Pareto-distributed ON and OFF periods, forms a self-similar data traffic. In the computer simulation, we consider three service classes representing real-time voice, real-time Variable Bit Rate (VBR) Markov Modulated Poisson Process (MMPP) video and data traffic. Numerical results indicate that the delay performance for heavy-tailed ON/OFF data traffic is well within the delay tolerance under light to medium loads.