Towards an AI-Enabled Signal Optimization Methodology in Vehicle Architecture Models

The increasing sophistication of functionalities in modern vehicles has led to significant growth in their architectural complexity. To manage this complexity, describing vehicle, system, and sub-system architecture is often done using the Systems Modeling Language (SysML). Signals serve as the primary means of interaction between different functionalities and systems within SysML models. In this paper, we present a GenAIenabled signal optimization methodology aimed at managing the complexity of system architecture models in vehicle development projects by finding and removing redundant architectural signals, and modifying existing ones to consolidate signal communication in that system architecture. This optimization leads to an easily scalable, and maintainable architecture that could support iterative development in agile environments. Ultimately, we aim through this optimization to reduce the cost and time required for the design and implementation of new functionalities in automotive system, as well as provide an example of the potential of GenAI in the field of systems modeling. This methodology has been developed through multiple iterations to improve its optimization capabilities and acceptance by system architects, and is integrated into an existing modeling tool (Cameo Systems Modeler). Finally, it is evaluated in the context of domain expert interviews.