A research submitted in partial fulfillment of the requirements for the degree of Master of Engineering, School of Engineering and Technology

Research Studies Project Report (M.Eng.) - Asian Institute of Technology, 1994

This study demonstrates the blocking probability sens1t1v1ty of the routedimensioning methods in the circuit switched loss system due to the measurement error of offered traffic parameters (mean and variance). The Equivalent Random Theory (ERT), which is based on the theoretical Erlang model, and the extended theory of ERT (EERT) cope with all kinds of the offered traffic. The other method is based on the empirical model; Normal Distribution Method (NDM). Error in traffic load measurements introduced by the method of observation arises from two principal sources: error caused by the limited duration of the measurement and error due to the scanning process. The confidence limit of the measured traffic mean and variance can be calculated. The variation in the traffic mean and variance causes the variation in the blocking probability too. Because the blocking probability is not linearly varied with the input parameters, the dimensioning of traffic route in order to achieve the level of grade of service is also effected. The four important parameters for traffic calculation are the offered traffic mean, offered traffic variance, number of circuits and blocking probability. In the same condition, the time blocking variation of NDM and call congestion variation of ERT-EERT have similar behavior. The sensitivities of the blocking probability from the variation of peaked traffic and smooth traffic due to the measurement error are in the same fashion if the network is designed to achieve the limited blocking probability for each type of traffic. The results are presented graphically. Various curve sets are made to demonstrate the influence of measurement errors on route dimensioning.