Trainers with a centralized architecture have inherent limitations on expandability, reconfigurability and ease of integration. The motivation for this project stems from the inability of the centralized architecture to address the complexity of individual, subteam, and team communication requirements for future trainers. With recent advances in commercial Implementations of local area network technologies, trainer systems can be designed to overcome these limitations and offer other advantages. This paper discloses the research, selection, integration and demonstration of a distributed bus architecture for trainer systems. First in the paper, an analysis of top-level trainer system requirements is given from which specific bus performance characteristics are derived. An in-depth discussion of the network selection process includes an explanation of the network selection criteria along with the specific characteristics of several comparable local area networks. Next, a description is given of the implementation of a commercial fiber optic ETHERNET network. The upper layers of network protocol software are presented. This bus system was demonstrated as a validation of research findings. Data throughput rates, derived from both theoretical predictions and laboratory measurements, are disclosed for typical trainer configurations. A performance model written in BASIC, which allows predictions of the bus performance based upon user input of trainer configuration, is described and will be distributed to interested parties.
Implementation of a Distributed Bus Architecture on a Trainer System
2 Views