Modern aircraft heavily depend on computer systems to carry out various tasks. From managing flight surfaces and engines to processing radar and imagery data and facilitating communication with other aircraft and ground stations, computers are involved in almost every aspect of an aircraft’s operation. These computer systems, known as Integrated Modular Avionics (IMA) systems, have long life cycles that span several decades and undergo regular updates. Despite this, a significant portion of the overall life cycle costs is determined very early in the life cycle, in the concept design phase. 

While the early concept stage provides the best opportunity to influence the design of the system and its future costs, it is also the stage where information about the system is most limited. During this early stage, selecting a suitable IMA platform configuration must ensure sufficient resources for the intended aircraft functionalities, particularly in computing and networking capabilities. Additionally, the decisions regarding safety and security measures must align with application requirements. However, this is a complex task due to conflicting requirements, necessitating compromises, and the limited information available at this early stage. 

This thesis explores the analysis and generation of avionic architecture configurations during the concept stage, addressing the problem on two fronts. The first focuses on verifying whether a chosen IMA platform configuration provides sufficient resources to ensure timely communication for a specified set of avionic applications. The second centers on exploring the conceptual design space to find IMA platform configurations aligned with computing, networking, fault-tolerance, and security application needs. 

To contribute to the problem’s verification aspect, this thesis introduces two high-level abstractions, namely timed automata and a domain-specific model based on Unified Modelling Languages (UML), to model IMA systems at the concept stage. These are designed to capture inter-process message ex-changes within networked IMA platforms. Additionally, we propose a workflow and a supporting tool explicitly designed to translate our proposed model into a network calculus model for further analysis. The approach’s practicality and scalability are showcased through its application to an avionics use case. 

In exploring conceptual design space, this thesis proposes NetGAP, a domain-specific method in which interconnection patterns in generic networked system topologies are represented as graph grammars. Combined with Monte Carlo Tree Search and genetic algorithms, these grammars are used to navigate the solution space and generate candidate IMA platform configurations tailored to the requirements of an envisaged application. Through application to an avionics use case, NetGAP is shown to be scalable and suitable for different types of requirements. To further expedite the process, NetGAP has evolved into NeuralGAP. The latter employs graph neural networks to assess network topology compatibility with the target application, accelerating the concept exploration and improving its results.