Performance of Aloha channels under self-similar input

Abstract

The authors perform a trace-driven simulation experiment to analyse the performance of the Aloha channel under self-similar input. The real traffic trace is obtained by recording four days worth of data in the UCB campus network. The Aloha channel is assumed to carry delay-sensitive, non-bandwidth-intensive traffic, such as that generated by a Telnet client, which presents self-similarity features. The authors' findings show a significant increase in normalised throughput and a drop in packet collision probability in comparison with the Poissonian case.