Configurable Depth Analysis: Create a meta analysis to statically analyze a system in different depths

Typ der Arbeit: Masterarbeit
Status der Arbeit: abgeschlossen
Projekte: AHA
Betreuer: Gerion Entrup, Daniel Lohmann
Ende der Arbeit: 13. Jun 2023

ARA is a whole-system optimizer. As such, it consumes a (real-time) system, analyzes it to gain some knowledge and performs various optimizations on that base. For knowledge retrieval it supports mainly this analyses:

The Static Instance Analysis (SIA) and Interaction Analysis (INA): Two flow insensitive analyses to statically retrieve instance and interaction knowledge. They are resource friendly and fast, but imprecise.
The System-State Enumeration (SSE): A flow sensitive analysis that builds a graph of all possible system states. This analysis is precise, but slow and costly.

Embedded systems often are dividable in parts of different criticalities (mixed criticality). In the simplest case, this results in a small but very important part and big but less important part. Your goal with this thesis is to bring this classification of criticalities to the analyses: The important part should be analyzed with the most precise analysis, the SSE. The unimportant part should be analyzed with the more imprecise but faster SIA and INA.

This task has several challenges: The analyses need to be adjusted from the handling of the whole system to only a specific part. You need to find an appropriate split point into "important" and "unimportant" and you must handle cross-domain interactions (what happens if an action in the "important" domain triggers something in the "unimportant" domain and vice versa?). All of that should be implemented within ARA and proven by a minimal working example.

Topics covered by this thesis:
static analysis, AUTOSAR, Python, C++, C

Configurable Depth Analysis: Create a meta analysis to statically analyze a system in different depths

Further Reading