Traffic Pattern-Based Comparison of Bulk Image Request Response Times for a Virtual Distributed Laboratory

摘要

Various agencies throughout the Department of Defense possess intelligence imagery and electrooptical signature data required by researchers in the field of automatic target recognition (ATR). The Air Force Research Laboratory, Sensors Directorate, as been tasked with creating a virtual distributed laboratory (VDL) which will make this data available to ATR researchers via high speed networks such as the defense research and engineering network (DREN). For this research, a model for simulating potential operational network configurations and collaboration scenarios was developed and implemented using OPNET. The results of the simulations were analyzed using statistical methods to determine the impact on performance of network configuration, connection speed, server capability, and data size. Connection speed proved to be the ultimate limiting factor on system performance, but statistical insights regarding usage patterns and file sizes are drawn from the results as well. This research provides VDL designers with performance trend data an enhances the design process by providing insight into how design decisions will affect future network performance.