Overview of the research program

Many modern industrial systems fall in the realm of Cyber-Physical Systems (CPS) because of the tight interaction between computation, communication and control elements (the cyber part), and physical processes such as motion, heating/cooling, vibration, wear and tear (the physical part) within these systems.

Traditional design methods involve multiple, often isolated, design phases involving different disciplines (mechanical, electrical, control and software engineering). Requirements related to cost, quality and reliability enforce designs with over-provisioning of platform resources (computation, communication, memory) by large quality and safety margins at each phase to be able to fulfill system-level requirements in the worst-case scenarios. Due to increasingly stringent cost and resource constraints these overly conservative designs are no longer sustainable. As a consequence, there is an urgent need for integrative design trajectories that allow for tradeoffs between cost, quality and reliability coping with the tight coordination between the cyber and the physical components. This gives rise to the need for models that accurately capture the interaction between various components (e.g., software, electronics, mechanics, algorithms, power, energy, etc.) and novel design methods that exploit the artifacts of the underlying platforms. This requires the integration of a number of scientific disciplines that have predominantly developed independently.

oCPS research methodology

High-level view of CPS consisting of three abstractions: The platform is consisting of hardware (computation, communication and memory resources) and software (tasks, messages, mapping and scheduling). The physical system represents the physical processes in the particular application at hand (e.g., automotive/energy/robotics/healthcare systems). Functionality has to take into account the platform (cyber) and the physics (physical) as well as non-functional requirements such as cost, quality and reliability (optimization goals). The functionality is realized by model-driven synthesis of platform components (software and/or hardware). Sensors and actuators establish the interaction between functionality and the physical system. Models of the physics, the functionality and the platform at a variety of abstraction levels play a key role.


Research methodology: The key scientific objective of the oCPS program is to enable the design of a new generation of cost-effective, quality-driven and reliable CPS by developing model-driven design methods that capture the interaction between different models at various design layers, that take into account physical constraints and processes, and that introduce platform-awareness at all levels. Starting from the societal need for high-quality, low-cost, reliable products in a variety of CPS application areas, we identified a number of key technical challenges. The technical challenges are translated into five research lines (RLs) that will be investigated by fifteen Early Stage Researchers in the oCPS program.

Research lines

RL1: Multi-Domain Modelling and Architecting

It is essential for the successful development of CPS to carefully exploit the interaction among models in various phases of the design cycle towards meeting functional and non-functional requirements . This requires integrative system-level approaches across the phased development cycle – common in domains like industrial automation. On the one hand, RL1 addresses choosing the right level of detail in models at each phase of design. On the other hand, the dependencies among various physical and cyber (platform) components are utilized in architecting models across phases and domains.

Projects: (click on each project to get more information)

(ESR11) Reliable design of cyber physical systems based on partial and heterogenous models
(ESR13) Improving CPS design efficiency and performance through multi-view modeling
(ESR5) n.a.

RL2: Resource-Constrained Design

CPS domains like healthcare, automotive, low-power imaging, and robotics have stringent constraints on platform resources (computation, communication, memory) and non-functional requirements (power, performance, reliability). While resource sharing among applications is a potential solution, inter-application interference poses challenges both in guaranteeing correct functionality and in satisfying constraints. The RL will focus on development of CPS-domain-specific analysis and synthesis of platform components to optimize both functional and non-functional aspects and to meet constraints.RL4: Cost-effective and Reliable Synthesis of Control Software

Projects: (click on each project to get more information)

(ESR4)Low power vision architectures using approximate computing
(ESR9) Timing Analysis for Cyber-Physical Systems
(ESR14) Software Synthesis for Power/Energy Constrained Applications 

RL3: Platform-Aware Control Systems

Maximization of performance and minimization of implementation cost (platform resources) are conflicting when realizing control algorithms on embedded platforms. This has obvious relevance in CPS domains such as automotive, robotics, energy systems and healthcare. RL3 aims to extend classical model-driven control methods to take into account platform details, e.g., platform precision, scheduling policies and memory architectures.

Projects (click on each project to get more information):

(ESR1) Co-design of Control and Streaming Applications Considering Tradeoff between Quality-of-Control and Quality-of-Service
(ESR15) Industrial Embedded Model Predictive Control (MPC)

RL4: Cost-effective and Reliable Synthesis of Control Software

Improving reliability and minimizing cost of implementation (e.g., platform resources, energy, power) is yet another conflicting pair of objectives which has particular relevance in safety-critical CPS application areas. RL4 focuses on development of cost-effective synthesis of reliable control software meeting functional and non-functional requirements.

Projects: (click on each project to get more information)

(ESR3) Model-based Engineering of Supervisory Control for Cyber-Physical Systems
(ESR7) Compositional Abstraction of Networks of Stochastic Hybrid Systems
(ESR10) n.a.

RL5: Distributed Coordination

CPS are inherently distributed. Multi-core processors form the heart of many components; components are put together into systems and systems into (networked) systems-of-systems. Examples of the latter are domains dealing with distributed systems like energy distribution and vehicle platooning. RL5 will address challenges involving coordination among components of CPS.

Projects: (click on each project to get more information)

(ESR6) Communication-aware analysis and design of control algorithms for platooning
(ESR12) Distributed Control of Cyber-Physical Systems with Platooning Applications
(ESR8) Large-scale Cyber-Physical Systems with Resource Constraints
(ESR2) Resource-aware control: An Approximate Dynamic Programming Approach