Trustworthy Adaptive Computing

The seamless integration of intelligent embedded systems into our everyday life raises urgent questions about their trustworthiness. Such trustworthiness can be characterized along a system’s functional correctness (adherence to the intended functionality), availability (readiness for service), reliability & resilience (continuity of service), safety & security (prevention of catastrophic consequences and misuse), integrity (absence of improper system alteration), maintainability (the ability for easy maintenance), as well as the ability to adapt to changes in its environment (including other systems and humans it interacts with), status (internal faults and aging), and assignments (the tasks it shall perform).

Important questions in our research are

how to ensure the adherence to the intended function also when design assumptions do no longer hold,
how to make a system’s observed data efficiently and dependably available in all the places required for the system to be able to make the right decisions, or
how to ensure an embedded system’s security also in the post-quantum computing era.

Answering these questions with our research on "Trustworthy Adaptive Computing", we strive to enable trustworthy embedded systems – efficient, functionally correct, resilient, safe and secure.

Research Foci

Efficient trustworthy computation
- Advanced signal processing with embedded computations - make observed data efficiently and securely available where required
- Post-quantum cryptography (PQC) – implementing efficient and secure PQC algorithms and hardware
Trustworthy AI
- Fusing model-based/symbolic AI with learning-based/sub-symbolic AI and formal methods to enhance explainability and verifiability
- Resilience to disturbances (faults, uncontrollable external events), uncertainty (resulting from, e.g., limited observability), and novelty (changes in the environment)

Research Competencies

Algorithms/AI
- time series analysis, e.g., predictive maintenance
- energy-efficient and high-performance (signal processing) algorithms
- Model-based diagnosis and reasoning
Security, Verification & Validation
- Temporal logics, formal verification and methods, automata-based concepts
- Side-channel vulnerability analysis of electronics-based systems
Software Technology
- Software Engineering: compiler construction, reverse engineering
- FPGA programming

Applications

Novel hardware concepts for ultrasonic transducers, signal generators and signal processing in the analog and digital domain
Intelligent battery management system for battery sharing of e-bikes
Reverse-engineering and side-channel attacks on AI-Hardware
Explainable AI through automata learning
Distributed learning and data management

Your contact person

Dr. Willibald Krenn

Head of Research Unit Trustworthy Adaptive Computing

e-mail: contact@silicon-austria.com