PROVISIONING FOR BANDWIDTH SHARING AND EXCHANGE

Customers of bandwidth services can be divided into two distinct groups: those customers requesting bandwidth for the future and those desiring bandwidth immediately. We develop a dynamic network provisioning methodology that minimally satisfies the QoS (blocking probability) requirements for the ’on-demand’ customers. Our method is sufficiently general and captures time varying trends in the demand for services as well as different bandwidth requests for the multiple classes of customers. This allows a network provider to be efficient in reserving excess bandwidth for forward contracts. Asymptotic results and bounds for the Erlang loss system are invoked to obtain simple approximate solutions to this bandwidth provisioning problem.