We are pleased to welcome Dr. Tatara as keynote.
Full agenda available => See next tab
The STEADINESS workshop is organised as part of the TechNexus Programme. This is an initiative to support developing a community focusing on technology integration for complex and critical systems and as you will know, this is a cross-disciplinary and challenging topic to address. There is a particular need to have such a community at present, given the acceleration of advances in specific technology domains and the highly interconnected nature we see arising.
STEADINESS encourages participation from system-level communities including architecting and modelling, orchestration technologies such as safety, security, performance and useability, validation and verification and product life-cycle. Product-side and market influencers, such as from product-lines and policy making are also invited to participate. Topics should include bridging advances between these domains, or to functional level domains, including supports via enabling technologies like edge\cloud continuum, IoT, Systems of Systems, AI, software and hardware.
All presentations will be made publicly available to a wide audience via HiPEAC. Furthermore, based on STEADINESS and FORECAST presentations and discussions, a white paper will be produced, which will be published openly to support research applications and recommendations to future research calls (last year’s white paper can be found here: https://zenodo.org/records/14920027). We plan to have a similar white paper this time.
Complex and critical systems relate to a large proportion of national infrastructure (transport, health, manufacturing, space, energy, etc) and the target for many of the new technologies emerging. However, without a dedicated focus on facilitating the integration and uptake of new technologies by the complex systems, such as in our national infrastructure – we limit to a large extent our capacity to use the new technologies. Additionally the high level of interconnectivity brings new risks, particularly relating to safety and security, without comprehensive approaches. On the other hand, by building an interaction point and supporting such research, we help industry to remain competitive by easier technology uptake, which also supports home grown technologies to be used nationally (sovereignty).
All parties of the dependable systems ecosystem are really needed to broach this challenge, the diverse technology contributors, production line managers, policy makers and end users. The TechNexus Programme is currently coordinated by Charles Robinson (Thales).
At HiPEAC, the TechNexus Programme organises two workshops this year, bringing the the system engineering domains communities (STEADINESS, Monday, 26 January 2026) and the functional properties communities together (FORECAST, Tuesday, 27 January 2026).
ORGANIZERS
Session chaired by:
- Géza Kulcsar (PTC, Hungary)
- Bernd Westphal (DLR, Germany)
KEYNOTE SPEAKER
Pleased to welcome Dr Marek Tatara of Gdańsk University of Technology as a keynote speaker.
Trust but verify - engineering reliability into agentic systems
Overview:
The definition of a system is evolving. We are entering an era where software does not just compute - it reasons, plans, and collaborates. However, for high-performance and critical applications, reasoning without reliability is a liability. We are moving from explicit instruction execution to goal-oriented orchestration, where the connection between input and output is no longer purely deterministic. This talk explores the architectural complexity of building reliable agentic teams and the engineering measures required to bridge the gap between stochastic reasoning and deterministic execution.
The friction between the probabilistic nature of Large Language Models and the deterministic requirements of classical software engineering is addressed. From this perspective, the orchestration of agent teams and the specific utility of the Model Context Protocol (MCP) are analyzed in the context of creating rigid interfaces for agentic behavior. The discussion extends to the Agent-to-Agent (A2A) protocol, with an examination of the associated risks related to inter-agentic communication. The session concludes by presenting an approach for quality assurance, consisting of a multi-level testing strategy that combines the formal verification of tools with the behavioral evaluation of autonomous agents and teams.
WORKSHOP FOCUS & TOPICS OF INTEREST
The scope of complex and critical systems is extending significantly in many directions and application domains due to advances in embedded technologies and increased interconnectivity made possible by mobile and ad-hoc wireless communication. Maintaining a top-down system view is vital for such systems, requiring support of aggregating technologies and those that treat the system as a whole, to guarantee dependability, performance and to optimise component integration/interaction. The engineering challenges are high to manage this complexity, and there is still a gap between the model-based design approaches and the software frameworks to implement trustworthy systems. To make things even more challenging, modern ICT systems must not only provide the required end-to-end functionality in a robust and dependable way, but they must also be environmentally neutral and exploit green energy in a smart way.
This workshop seeks theoretical as well as practical contributions in the area of complex and critical systems. In particular, presentations covering topics bridging from system engineering domains are welcome such as the following:
System Modelling
- Model-Driven Engineering methodologies and tools.
- Domain Specific Modelling Languages from different industrial domains, targeting CPS preserving the dependability and performance of the system.
- Optimised software synthesis methods, with special interest on those targeting parallel and heterogeneous execution.
- Software-Defined Everything, eg. Vehicles, Robotics and cyber-physical systems
Validation and Verification
- Methodologies, techniques and tools for functional and system properties.
- Timing verification tools and techniques, especially executed in multicore and heterogeneous platforms, statistical and probabilistic approaches for timing analysis, and artificial intelligence techniques for timing bounds predictions.
- New dependability assessment techniques for complex systems based on analysis, testing or simulation. Early dependability assessment methodologies at different abstraction levels (software, microarchitecture, and physical design).
Coupling of safety-security-performance at system level
- Co-design approaches
- Approaches enabling automatic comparison of these properties (by value rather than general if-else pre-defined
- Knock-on effects analysis of safety-security-performance.
- Advances on mediation capabilities between standards\certification
Human-Machine Interaction:
- Advanced multimodal interfaces for complex system control
- Human factors in safety-critical system design
Resilience and Fault Recovery:
- Predictive maintenance and fault detection
- Graceful degradation strategies for critical functionalities
- Cross-domain fault propagation analysis and mitigation
Networking and Security
- Network Programmability and In-networking computing, e.g SDN, P4, DPUs
- Time Sensitive Networking
- Confidential Computing
Bridging challenges in systems engineering, being supported by enabling technologies, could include for instance:
- Artificial Intelligence
- AI for dependable and critical systems and CPS.
- AI for system and application management and configuration in continuum systems.
- AI for distributed/swarm intelligence.
- Edge-centric Federated Learning and Inference for CPS.
- Robust and Explainable AI for CPS
- Systems of Systems & Continuum Systems
- Approaches that consider a CPS not as a tightly-coupled system but as a complex system that is built by combining several other systems, possibly managed by different authorities.
- Methods supporting the development of distributed applications that can span the entire continuum, from the cloud to the edge and even on standalone IoT devices.
- System software supporting application deployment and execution across the continuum, taking into account the application’s QoS requirements and supporting runtime adaptation.
- Languages, Compilers and Development Tools for Heterogeneous Complex and Critical Systems
- Extensions for parallel programming models and languages from syntax, semantic and implementation perspectives, to adapt parallel frameworks to better express the dependability of CPS, particularly considering heterogenous platforms.
- Compiler methods to optimise the development and execution of complex platforms, including, but not restricting to, parallel and heterogeneous computing platforms.
- Tools that simplify the development of high-performance applications targeting heterogeneous computing platforms and accelerators such as GPUs, TPUs, FPGAs.
- LLMs for designing, writing and testing CPS software
PRESENTATION SUBMISSION INSTRUCTIONS
- The workshop welcomes presentations describing new ideas, work in progress or experience reports both from industry and academia. The presentations may include materials already published or presented in other venues.
- Presenters should submit a one-page extended abstract outlining their proposed 20-30 minute presentation.
- The submission should include a title, name of presenter(s) and the description of the proposed talk, including details of the technical bridging challenges they face with bridging from system level domains to others. There is no template, please send a PDF with all information.
- Please send the presentation proposals to : [Géza Kulcsar](mailto: [email protected]). There will be an acknowledgement e-mail.
ACKNOWLEDGEMENTS
This workshop has received support from the projects Arrowhead fPVN (fpvn.arrowhead.eu/fpvn-arrowhead/) grant agreement no. 101111977 & HiPEAC (https://www.hipeac.net) grant agreement no. 101069836.
These workshops are discussion based - a link provided here on the day to share notes.