A model relating measurement and forecast errors to the provisioning of direct final trunk groups

This paper describes a mathematical model of the provisioning of direct final trunk groups with forecasting and measurement errors. This model can be used to study the effects of applying standard trunking formulas to possibly inaccurate load forecasts. An important consideration in the process is the degree to which the trunk forecast is actually followed. This so-called provisioning policy is modeled parametrically to allow consideration of a range of strategies, from following the forecast precisely to complete reluctance to remove trunks when indicated by the trunk forecast. When load forecast errors are combined with a reluctance to remove trunks, there will be a net reserve capacity on the average, i.e., more trunks than would be needed if the loads were known exactly. Using the mathematical model, a set of curves known as Trunk Provisioning Operating Characteristics is calculated. These relate percentage of reserve capacity to service (as measured by the fraction of trunk groups with blocking exceeding 0.03). The accuracy of the estimate of the traffic load defines the curve on which one is constrained to operate. The degree of reluctance to remove trunks together with the traffic growth rate determines the operating point. Improved estimation accuracy corresponds to a more desirable operating characteristic. The accuracy of the forecast load estimate is influenced by many factors, such as data base errors (e.g., measuring the wrong quantity due to wiring or other problems), recording errors (eg., key punch errors), and projection ratio errors. This type of modeling may be useful both in evaluating the potential effects of proposed improvements in measurement or forecasting accuracy, and in studying the effects of changes in provisioning policy. A discussion is given of trunk provisioning process issues based on the viewpoint presented in this paper.