Master Thesis - 2014

In recent years, the number of requests to transfer large data files via large high-speed computer networks has been increasing rapidly. Typically, these requests are handled in a ``best effort'' manner, resulting in unpredictable completion times. In this study, I consider a model where a file transfer request must either be completed by a user-specified deadline or be rejected if the deadline cannot be satisfied. In this model, the assigned bandwidth should be efficient for reducing the number of rejected requests. In addition, for satisfying the request's deadline strictly, bandwidth-guaranteed network is assumed. In this model, we proposed a bandwidth scheduling method and a dynamic routing method. This method changes routes of file transfer requests dynamically, and reduced call-blocking rate by routing method considering remaining bandwidth and file size. However, assigned bandwidth is decided by the bottleneck link of the whole path. I discuss a routing method with a metric that can represent the congestion situation in links in the future and in the whole path. To consider the future congestion, a link metric based on approximated prediction of the assigned bandwidth is proposed. In addition, to consider the congestion in whole path, the number of requests sharing the link is also taken into account. Using these metrics, a method that aims to maximize the assigned bandwidth is proposed. Finally, I present simulation results that show its performance.