Optimal capacity planning in stochastic loss networks.

A large number of application areas involve resource allocation problems in which resources of different capabilities are used to provide service to various classes of customers at their arrival instants, otherwise the opportunity to serve the customer is lost. Stochastic loss networks are often used to capture the dynamics and uncertainty of this class of resource allocation problems. A wide variety of examples include applications in telephony and data networks, distributed computing and data centers, inventory control and manufacturing systems, and call and contact centers. Another emerging application area is workforce management where, e.g., an IT services company offers a collection of service products, each requiring a set of resources with certain capabilities. The customer demands for such IT service products are stochastic and the IT services company seeks to determine its per-class resource capacity levels in order to maximize its profits over the long run.