A Model Relating Measurement and Forecast Errors to the Provisioning of Direct Final Trunk Groups

Traffic measurements in the Bell System are used as the basis of those efforts aimed at planning an efficient network by providing appropriate quantities of trunking and switching equipment. They also form the basis of many efforts aimed at efficiently administering the network as well as serving as primary inputs for purposes of evaluating network performance. New traffic measurement systems typically bring with them a variety of benefits such as more accurate and detailed data and more automated and convenient collection and processing of the raw data, together with possible new uses for the data which these improvements allow. Of course, to prove in economically, these advantages must offset the costs associated with installing and operating the system. Clerical savings associated with data collection and processing represent a good example of a relatively easily quantifiable advantage. Other advantages are not so simply equated to dollar savings. To quantify the traffic-related benefits of improved measurements, models of the trunk provisioning process are required. We describe here a mathematical model of the provisioning of direct final trunk groups* with measurement and forecasting errors. An important consideration which ultimately determines the number of installed trunks is the degree to which the trunk forecast is actually followed. This socalled provisioning policy is modeled parametrically to allow consideration of a range of possibilities, from following the forecast precisely to complete reluctance to remove trunks when indicated by the trunk forecast.