Performance implications of Pareto-distributed cell residual time in distributed admission control scheme (DACS)

While many studies of admission control schemes assume exponentially distributed cell residual time, field measurements have shown that cell residual time, in reality, follows a heavy-tailed distribution; in particular, a Pareto distribution. These measurements were made in Wireless Andrew, an enterprise-wide broadband microcellular wireless network at Carnegie Mellon University. In this paper, we investigate the impact of Pareto distributed cell residual time on the admission threshold of the distributed admission control scheme (DACS), which is one of the well-known admission control schemes. Our results show that under Pareto-distributed cell residual time, a system that implements DACS can accommodate or "sign on" more simultaneous connections than the exponentially distributed case.