|
|
|
|
|
|
|
|
|
|
|
|
|
Energy Efficiency | |
Contact Person: | |
| We aim at the development of a comprehensive concept for energy efficiency, involving all layers. This involves physical nodes, cooling of nodes, networking hardware, communication protocols, and finally the services themselves that running on the nodes. |
| |
| |
 | |
IT-Security | |
Contact Person: | |
| Overlay networks, most recently the peer-to-peer (P2P) paradigm, have added new services to networking. While overlay networks enhance communication efficiency and robustness of popular service applications, trust into all parties is assumed. Secure and fair operations have been neglected in many approaches. Our research aims to 1. strengthen cooperative behavior in P2P-based networks 2. enable secure and private, unobservable operations. Integrity-based computing enforces data integrity in distributed systems, and therefore e.g. Trusted Platform Models can provide a fundament for secure cooperation. Peer-to-peer networks distribute their service at the price of additional attack possibilities. Malicious behavior can appear on various levels, from Denial-of-Service attacks utilizing community resources to unobservable misbehavior and undermining policies. The lack of peer-to-peer networks of monitoring, accounting, and enforcing security policies becomes a thread of future networks. The emerging trend of network privacy strengthens data protection of individuals. Anonymity systems protect the relationship between identities and events, e.g. content requests, mails, other traces of network activity. Captured communication relationships can offend privacy of individuals. Research in this field must enhance the usability of solutions. Recent exposures showed that simple approaches of pseudonym data do fail and reveal sensible data of individuals. Our research combines network privacy with mechanisms for secure cooperation. This is beneficial since unobservability is only reachable with a reasonable anonymity set, i.e. multiple cooperation parties. |
|
|
|
|
| |
Virtualization | |
Contact Person: | |
|
The management of virtualized resources is a main research field of our research group. Resources are virtualized in various different contexts:
In data centers, e.g., server hardware is virtualized by using system virtualization methods. This virtualization enables the creation of virtual machines that can be used similar to physical servers. Virtual machines can be created, destroyed, copied or moved from server to server. Furthermore, more than a single virtual machine can be hosted by a physical server. This is used to achieve a consolidation of virtual machines on a small number of physical servers, which saves hardware and energy. This kind of virtualization and consolidation can also be extended to office environments and home environments.
Another field that virtualizes resources is, e.g., network virtualization. Services, routers, and links are virtualized in order to form virtual networks with highly interesting properties. These networks are a possible approach towards a Future Generation Internet and provide important features like quality-of-service, high availability of services, or resilience against disasters. Virtualization can also be applied to Future Home Environments, to enable a fair sharing of resources among users while reducing energy consumption.
All these (and other) applications of virtualization require a management of virtual resources which is highly dynamic, mostly autonomic and often based on a decentralized approach.
Peer-to-Peer (P2P) and Overlay networks, which are a special variation of network virtualization, are also a research field of our research group. This involves the analysis of different P2P paradigms and their traffic patterns (signaling traffic and data traffic), cross-layer optimization, and mobility in P2P-overlays (Mobile P2P). |
|
|
|
|
| |
Selforganization | |
Contact Person: | |
| |
|
|
Related Events: | |
| |
Performance Modeling | |
Contact Person: | |
| One of the most common ways to evaluate the performance is to measure it. However, direct measurements at real systems are only possible if the system is already available. Additionally, doing optimizations based on measurements often involves costly trial-and-error cycles. Therefore, our evaluations are based on formal mathematical models of the real system instead. Here, we mainly focus on analytic closed-form solutions, iterative numerical methods, and discrete-event simulation techniques for models based on stochastic processes, like Markov chains, as well as high-level model description methods, like queueing networks and stochastic Petri nets. Especially for complex and self-organizing systems, we also employ models based on differential equations well-known in Physics and Biology research. |
|
|
| News |
|---|
| May 17, 2013 |
| April 25, 2013 |
| January 31, 2011 |
News at Department News at university Events at university
Home
People
Research
