Dynamic Bandwidth Allocation for Requests with Classes and Priorities in Preemptive Distributed Networks

In today\u27s general-purpose networked computing environments, both commercial and military, in addition to providing connectivity, providing quality of service (QoS) to users is a major concern. One of the major QoS parameters in this context is bandwidth. This research focuses mainly on the intelligent allocation of bandwidth to requests in oversubscribed networks such that, some measure of worth associated with the satisfied requests is optimized. In this work, heterogeneous networks with preemptive capabilities have been considered. A preemptive network is capable of interrupting a request that is currently in progress to service a request that is considered to be more important to the system by some metric. To provide a sound theoretical foundation for grouping requests into different categories, a class and priority based mechanism is described such that, all requests belonging to a higher class are to be satisfied before any request(s) of a lower class are considered. Within a given class, Wher differentiation among requests is achieved by assigning different priority levels to the requests depending on their relative importance. Scheduling heuristics have been developed to determine the bandwidth allocations to the requests depending on their class, relative worths of their priority levels, and the network capacities available for that period. These heuristics allocate network bandwidth to requests of two types: 1) session type - requesting a fixed amount of bandwidth for a given interval of time, and 2) data type - requesting a data item of given size with an earliest available time and fixed deadline. The heuristics used for the intelligent allocation of bandwidth try to maximize the weighted priority (or worth) of the satisfied requests in the most important class, then the second most important class, and so on, down to the least important class. Simulation experiments have been conducted to quantify the performance of the scheduling algorithms in different scenarios