DC13 Project: Trustworthy and Reliable Cyber-physical Systems


PhD Supervisor: prof. Volker Stolz (HVL), co-supervisor: Marcin Fojcik (HVL); Auxiliary supervisors: Prof. Shen Yin (NTNU); Francesco Piccialli (UNINA), Dariusz Mrozek (SUT); R&D cooperation: AIUT


Objectives: Distributed, dynamic, heterogeneous systems that integrate sensor networks, the IoT and mobile devices use low-power, proximity-based broadcast communication such as BTLE and UWB to collectively solve (computational) tasks. These tasks might involve AI-based techniques that are only available on a subset of the participating devices. Aggregate computing (AC) provides a unifying foundation for developing these systems, avoiding a centralised server and obtaining deployment on multiple hardware platforms. In the future, these systems will not only use immediate sensor-data but will also employ local machine learning-techniques. Within heterogeneous systems, the challenges are that: (i) resource-constrained devices will want to off-load classification to more powerful nodes nearby; (ii) the AI-based classifiers that are used will need to be updated and (iii) trust and resiliency will be established through consensus in each neighbourhood of devices. The FCPP library is a viable starting point that has already been validated for embedded devices, mobile phones and includes a (graphical) simulation and analysis framework.

The primary objective of DC13 project will be to develop a functioning platform that integrates AI-based techniques into an existing heterogeneous computing paradigm. Their secondary objective will be to develop techniques that ensure trustworthiness and to develop methods and tools to analyse and quantify the various performance parameters of the software on this platform such as degree of privacy, reliability and energy consumption.


Expected Results:  (i) implementing an intermediate self-organising layer in an AC that will enable resource-constrained device to discover the capabilities of their computationally more powerful neighbours with whom they can collaboratively solve computational tasks (T5.1); (ii) extending an AC with constructs that will use ML-based classifiers and integrate the sensor fusion and models from T2.1 & T2.3 into an AC deployment; (iii) implementing resilient, energy-aware anomaly detection based on the above (T5.2); (iv) designing and executing simulations and running their physical counterparts, conducting experiments to evaluate the performance aspects such as reliability/resilience and energy consumption, thereby contributing to T3.3; (v) releasing the libraries and example applications as open source, thereby contributing to WP6.


Applied research: Prototypes will be developed for and evaluated on AIUT's AGV-platform, HVL's Smart Software Systems IoT-lab.


Planned secondments: NTNU(4 months); UNINA(4 months); SUT (4 months)                   


Enrolment in Doctoral degree: HVL


Recruitment of Doctoral Candidate: https://euraxess.ec.europa.eu/jobs/298674