In telecommunications, the demand is a key data that drives network planning. Because of customers' movements and introduction of new services and products in the present competitive markets, the demand exhibits considerable variability. To deal with this uncertainty, we consider capacity assignment problems in telecommunications in the framework of robust optimization. Since in some situations different possible uncertainty sets may exist, we propose a generalization of these models to cope with a finite number of plausible uncertainty sets. With each uncertainty set is associated a weight that determine its relative importance or worth against another. We propose a decomposition scheme based on cutting plane methods. Some preliminary computational experiments indicate that the Elzinga-Moore cutting plane method can be a valuable choice.

An ATM switch has to process several services that require different quality of service (QoS) and resources. Thus, a uniform switch processing may lead to network performance degradation Thus adapting the switch processing to each traffic service is an important issue that will optimize the network resources. We present here a methodology to design an ATM switch. Our approach is based on switch processing adaptation according to the traffic behavior. Based on simulation techniques some experiments are carried out on different traffic behavior to identify the most suitable switch processing.