On the problem of applicability of synthetic data in testing intelligent transport systems

Intelligent Transport Systems (ITS) are now being implemented to ensure optimal and safe road traffic. Increasingly, these systems use artificial intelligence to obtain characteristics about traffic flows. The number of sensors and transducers is increasing dramatically, resulting in higher loads on ITSs. Therefore, it is necessary to develop distributed monitoring systems with scalability and fault tolerance in mind. However, extensive testing is required before implementation. It is not possible to fully conduct such testing on real data due to various factors. Therefore, this paper proposes a tool for generating synthetic traffic flow data with subject matter specificity. The generation system is designed to be integrated into different systems, which will allow different ITS vendors to use it. This service fulfils the scalability requirements and is close to real data. The study proposes a scalable architecture of intelligent transport subsystem that fulfils the requirements of scalability and fault tolerance. As part of this work, a testbed is assembled and the proposed architecture is tested through the developed service of synthetic traffic flow state data generation.