Providing different levels of network availability in high-speed networks

One of the desired properties of high-speed networks is the ability to provide different levels of grade of service (GOS) to various types of traffic. To achieve this property, each call is assigned a virtual circuit or connection. The network treats each connection and the data packets flowing in it differently according to the GOS parameters specified at the time the connection is established. During nonstationary network conditions when the network is under stress (for example due to link/node failures or during peak hours), it is desirable to provide higher network availability for calls of greater value (or higher priority) than others. This may require that ongoing low-priority calls be preempted to make room for new or rerouted high-priority calls. Call preemption when coupled with the capability to reroute calls (preempted due to preemption or link/node failures) creates a function whose behavior is an interesting phenomenon. The authors investigate the call preemption problem, introduce a simple connection-oriented call control protocol, present an efficient algorithm for selecting calls to be preempted, and present a simulation study of call preemption