Distributed, Embedded and Intelligent Systems

Distributed, Embedded and Intelligent Systems is one of  three research units at the Department of Computer Science. The unit covers real-time and distributed systems, networks, formalisms for the description and analysis of computer systems as well as tools for verification and validation.

The unit’s research concerns modelling, analysis and realization of computer programs with emphasis on distributed, embedded and intelligent systems. This includes the following areas:

• Semantic theories for modelling the behavior of computer programs and systems
• Design, implementation and models for analysis and construction of distributed, embedded and intelligent systems
• Algorithms, methods and tools for verification, and validation of programs and systems
• Probabilistic models and algorithms for intelligent decision making and machine learning

In 2001, the Center for Integrated Software Systems (CISS) was established in collaboration with the Department of Electronic Systems at Aalborg University. The focus of CISS is to create an industry-oriented research center of excellence with a distinctively visible profile within software construction and embedded and intelligent systems, making use of and creating new research within areas already existing in the unit.

Currently the unit is coordinating several major projects including the Innovation Center DiCyPS (Data-Intensive Cyber Physical Systems), the Sino-Danish Center of Excellence IDEA4CPS, and the ERC Advanced Grant project LASSO (Learning, Analysis, Synthesis and Optimization of Cyber Physical Systems).

Each of the four research directions mentioned above constitutes an area in its own right. Moreover, the areas are interrelated in a number of ways:

Semantic models offer important guidelines for development of languages and paradigms for distributed systems; semantic models are necessary prerequisites for development of verification and inference algorithms and tools; the development of validation and inference tools provides new insight into the underlying models on one hand, and are applied in environments for the construction and analysis of distributed, embedded and intelligent systems on the other; the usage of distributed, embedded and intelligent software in  increasingly complex systems provides valuable insight into the strengths and weaknesses of existing models, algorithms and tools, and serve as inspiration for development of new ones. Machine learning is central for data-driven development of intelligent software.  This complementing and completes the units focus on model-driven development.

The unit’s current research includes the following activities:

Foundational & logical THEORY

Semantic theories and meta-theories for concurrent processes and their logical properties. Semantic theories for processes which communication topology changes dynamically, including security protocols. Study of quantitative extensions such as hybrid, real-time and probabilistic processes, compositionality and behavioral metrics.

VERIFICATION AND VALIDATION

Development and implementation of data structures, algorithms and tools for model-checking, testing, performance analysis and synthesis for embedded systems focusing on real-time, probabilistic and hybrid aspects. Applications to communication protocols, control programs and planning and scheduling.

EMBEDDED SYSTEMS METHODOLOGY

Methodologies for specification, analysis and testing of embedded systems. Modelling and analysis of industrial case studies. Component-based development of embedded and hybrid systems.

NETWORKS & OPERATING SYSTEMS

Analysis and construction of services and protocols for computer networks, including grid computing, high-performance computing and software defined networks. Real-time and embedded operating systems.

PROBALISTIC GRAPHICAL MODELS

Developing efficient design and inference methods for graphical models; in particular, frameworks for representing and solving complex decision problems under uncertainty.

MACHINE LEARNING

Learning from data for knowledge discovery and design of intelligent systems. The focus is on the use of probabilistic and statistical methods, as well as the use of logic-based methods for modeling complex, structured data.