Quality of modelling vs quality of results: Case study of wireless network performance

Performance modelling is a powerful technique to predict the Quality of Service (QoS) of ICT systems. However the quality of the obtained results is not strongly related to the accuracy of the model. This paper presents a case study about the QoS of cellular wireless networks in which building a more detailed model does not imply a significant improvement of the obtained results. We will see that in many cases the effort spent in preparing a more detailed QoS model does not justify the improvement in the performance estimations. To justify this assertion we will analyze the QoS of a cellular wireless network by means of several models with different degrees of approximation to the real world and we will compare the results. Also we will use one of these models to improve the performance of this type of network by changing the call admission policy.