Research topics of the chair "Design of Cyber-Physical Systems" (Prof. Dr. Grimm)
The chair of "Design of Cyber-Physical Systems" deals with design and design methodology of cyber-physical systems. We focus three research topics:
- Platforms for the internet of things (IoT)
- Tools and languages for development of Cyber-Physical Systems
- Validation, verification and self-awareness by symbolic models
HW/SW Platforms for the Internet of Things
In the internet of things, not only computers, but all kind of things are networked: light-bulbs (e.g. via ZigBee), health-sensors at the human's body, up to complex manufacturing systems. The challenge is to enable new services and values from this. To achieve this, appropriate platforms - hardware and software- are needed. Particular requirements are often a very low power consumption to enable battery or even harvester-driven operation over a long time, and interoperability of all kinds at a semantic -- not only physical! -- layer across different domains of application.
The chair has gained vast experience in these topics since 2006 in various EC and national funded projects. We have developed hardware platforms (SmartCoDe: 3D integrated SoC for energy management), and also software platforms for domain-crossing interoperability based on generic classification of services.
Some publications on these topics:
- TR-FSM: Transfisition Based Reconfigurable Finite State Machine (ACM, 2012) describes a re-configurable architecture that combines advantages of micro controllers and FPGA, in particular for ultra-low power platforms.
- In the book "Embedded Systems for Energy Management and Smart Appliances" (Springer, 2012) we give a broad overview of different topics; it appeared also in Chines language (Springer, 2015).
Model-based development of Cyber-Physical and Analog/Mixed-Signal (AMS) Systems
Over the last 20 years, the chair has continuously contributed to the evolvement of languages for modeling mixed discrete/continuous resp. analog/Mixed-Signal (AMS) systems: Initially, we contributed to the development of VHDL-AMS, since 2002 we contributed to the development and standardization of SystemC AMS (Accellera std, recently IEEE 1666.1).
Recent research focusses two issues:
Power Profiling - Objective is to trace the power consumption from physical components to its cause, often in complex and distributed software systems. This enables the power optimization of distributed HW/SW systems.
Model Based Systems Engineering - Objective is to translate a SysML model into SystemC (TLM, AMS) models of the communication- and software stacks, maybe running on top of an Instruction Set Simulator (ISS), and physical components into Modelica models. We in particular also target to generate the required co-simulation infrastructure in a way that allows domain-crossing round-trip engineering.
Some publications include:
- SystemC AMS Language Reference Manual (LRM) und User's Guide
- SystemC based Power Simulation of Wireless Sensor Networks
Verification, Robustness and Self-Awareness of Cyber-Physical Systems
Cyber-Physical Systems and the Internet of Things introduced a new complexity. Due to its high complexity, the occurrence of faults is not a seldom exception, the normal case to consider in daily operation. Hence, such systems are required to be inherently robust, resilient, or even "self-aware". As an example, consider an autonomously driving car with its sensors and all possible scenarios. Would you trust verification by some test cases known to manufacturers, based on abstract models? How many test are needed to guarantee safe and reliable operation?
At the chair of design of Cyber-Physical Systems we develop methods for validation and verification of such systems. Our approach is to describe "uncertainties" in an abstract way by symbols that represent non-deterministic branches, ranges, or probabilistic distributions or branches, and to evaluate the overall system dynamics by a comprehensive symbolic simulation. To create a simple kind of "self-awareness", we compress the representation of all possible dynamics and use it during run-time for diagnostics and generation of error-reactions for unforeseen errors.
An overview paper was presented at IMSTW 2015 (Paris).