Department of Computer Science 7
Research
Our research can be divided into the areas
quality-of-service (QoS)-oriented system design
and autonomic networking. In QoS-oriented system design
we focus on the systematic design of networked and embedded systems. The
vision is to provide model-based tool support over the whole engineering
cycle in a seamless way to yield systems optimized with respect to QoS issues.
Quality-of-service is understood as an umbrella covering various aspects such
as traditional performance measures (e.g., throughput of a network,
system response time, loss rate, etc.) and dependability measures
(e.g., reliability of a network, availability of a server system, etc.)
but also real time, energy efficiency, security, and economics.
For the QoS-oriented system design, modeling in its various aspects plays a
key role and is complemented by measurements and tests. Models are used for
system specification (the group is experienced in using SDL, MSC, and UML
and related tools), stochastic analysis and simulation (experience exists in
using queuing networks, stochastic Petri nets, UML models, and several
simulation tools), real-time analysis, HW/SW co-design (synthesis and
optimization starting from high-level models such as SDL and graph models)
and modular software engineering for mobile and reconfigurable components
(driven by UML models). Test cases are also generated from high-level models
and provide information about the functional and QoS-related system behavior.
Measurements are performed for selected experimental networks
(e.g., a cluster-based Web-server, a soccer robot team, a WLAN,
a sensor network) and are used to validate and calibrate the models.
The autonomic networking research primarily focuses on autonomous
sensor/actuator networks that consist of a large number of small and cheap
sensor nodes and a team of mobile robots. Sensor nodes contain a sensing unit,
a processor, energy supply, and a wireless communication adapter;
sensor networks can be used to collect, process, and communicate measurement
data. Such systems provide many design challenges on several abstraction
layers, including medium access control, ad-hoc routing, self-organization,
large-scale distributed information processing in the presence of severe
resource restrictions, dynamic topologies and frequent failures.
Mobile robots extend the possibilities of stationary sensor networks by the
abilities of locomotion and real interaction with the environment.
Important research questions in both stationary as well as mobile sensor
networks are QoS aspects such as energy efficiency, performance,
dependability, and security. We also investigate how concepts from cellular
and molecular biology (e.g., cellular signaling pathways) can be applied in
autonomous sensor/actuator networks. Finally, we work on new concepts for
network monitoring and intrusion detection in the area of network security.
