An investigation of models and heuristics for scheduling data requests in a distributed computing environment
Abstract
Current military exercises require the warfighters to be able to access information from diverse sources and be able to process this data while resident in a hostile environment. If the data is not available, or there are delays in obtaining the information lives can be lost as a consequence. The research explored here examines the ability of users to request information from diverse sources (both in terms of locality and data type) and receive this information before the users' deadlines expires. Two models have been developed and explored for the case when the network is oversubscribed and not all requests in the system can be satisfied. These models are based upon systems that have been or are currently under development. The first model allows for interacting with the network and creating a schedule for transferring the request information. Three multiple-source shortest-path algorithm based heuristics for finding a near-optimal schedule of data transfers are presented. Each of the heuristics can be used with one of four cost criteria developed. In addition, two different weightings for the relative importance of different priority levels are considered. The second model does not have explicit control of the network and the heuristics work with the network manager to setup information channels that allow users to receive information. The goal is to create a near-optimal set of information channels that will satisfy the users' requests. Four components are used to create a value function that is used to order the information channels for presentation to the network manager. In addition, two different weightings for the relatived importance of different priority levels are considered. Simulation studies for both models have been performed that evaluate the performance of the heuristics developed. For both cases, it is shown that the heuristics perform well compared to upper and lower bounds developed. These results can be used to develop techniques that can be deployed in actual military systems. Furthermore, the models and heuristics studied here can be applied to other domains, such as business applications that utilize the internet to obtain critical information.
Degree
Ph.D.
Advisors
Siegel, Purdue University.
Subject Area
Electrical engineering|Computer science
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.