M. Jurczyk and C. Sullivan (USA)
network management, bandwidth tracking, channel preemption, reservation system
This work investigates a bandwidth tracking algorithm in a version of a distributed heterogeneous data dissemination system called the Agile Information Control Environment (AICE). Military networking environments have to provide uniform flow of prioritized information with QoS guarantees across heterogeneous overloaded networks. AICE-like systems accomplish this by assigning worth to individual communication requests and allocating network resources in a way to optimize the overall worth of the system. To implement an efficient optimization process, the probability of setup or rejection of communication requests has to be derived without any knowledge of the actual topologies of the underlying networks. In this paper, a bandwidth tracking algorithm is therefore introduced that is able to predict the setup/rejection of a communication request with a high accuracy of up to 95% in preemptive networking environments by tracking the average spare bandwidth of end-to-end communication channels. The tracking algorithm uses elements of exponential adjustment coupled with a binary search. This enables the tracker to quickly learn about changes in the topology of the underlying networks or changes in the load on the networks. The tracker is compared to a gradient bandwidth prediction algorithm recently proposed for the Network Weather Service (NWS) that provides network performance forecasts in metacomputing environments. It is shown that the tracker is able to achieve an accuracy that is up to 40% higher than that of the NWS algorithm.
Important Links:
Go Back
