Simulation for Smartnet scheduling of asynchronous transfer mode virtual channels

Abstract

Critical to the success of all future battlefield commanders is the rapid retrieval of relevant, time sensitive information. Some of this information will be available locally while the remainder is stored in the United States. DARPA's Battlefield Awareness and Data Dissemination (BADD) program attempts to deliver heterogeneous data to the battlefield using Asynchronous Transfer Mode (ATM) protocol. ATM was originally designed to implement dynamic virtual channels over duplex, high-speed, high capacity fiber optic cabling. The problem addressed was to determine which algorithm best schedules calls on BADD's ATM network that uses static virtual channels over simplex, error prone, long delay, satellite links. Because the BADD project uses ATM in such an unusual way, and because of the need to determine a schedule for transmissions over the heterogeneous static channels, we modeled BADD using the state-of-the-art network simulation tool, Optimized Network Engineering Tools (OPNET). We determined several modifications that must be made to existing network simulators to allow them to model next-generation networks. Our simulation shows that a greedy algorithm yields a 53% decrease in the overall completion time and a 46% increase in average bit throughput over FIFO scheduling