Skip to Main Content
Summary form only given. Despite the enormous volume of telecommunications traffic data, which allows rich models to be realistically tested and estimation variance to be made 'as small as desired' in some contexts, there are many important gaps in the statistical wish-list for tele-traffic measurement and modeling. A selection of important current problems is in detail, including the apparently simple problem of confidence intervals on mean estimates, and the effects of traffic thinning on recent models of packet arrival processes. Each is described, their importance in the tele-traffic context explained, and progress towards solutions outlined.