Upgrading JavaCat: A Distributed System for Time-Shifted Air Traffic Scenario Generation

摘要

As technology advances in the transportation sector, safety systems are developed to aid operators and supervisors prevent potentially fatal accidents. To ensure these systems work, they must be rigorously tested with numerous scenarios to ensure the system is safe for production. However, these systems cannot be tested using real data due to the fact that operators and supervisors should not need the system on a regular basis; potentially fatal situations must be created for testing, which would put innocent civilians at risk. To overcome this obstacle, another system can generate lifelike scenarios for testing using simulations when provided with actual transportation data. This paper describes enhancements to the capabilities of an existing system that uses a genetic algorithm with the goal to create simulated situations for aircraft that can be used to test their conflict detection safety systems using timeshifted data, which would aid air-traffic controllers. The outcome is to determine the most efficient way to distribute such a system to allow for complex scenarios with massive amounts of input to run quickly. A developed software system in accordance with the research results to provide a more efficient distributed scenario generator is also presented and evaluated. The system is successful in its main goals of being deterministic, fault-tolerant, and scalable.