Our research philosophy can be described as “Theory that matters!”. Theoretical fundamental research results are tested using different approaches such as simulations, software in the loop, hardware in the loop, and testbeds. Important, implementations are not merely used to show feasibility, but rather to derive further relevant research questions by exposing theory to the real world.
Our overall research interest is in the field of the upcoming fifth generation of mobile communication systems named 5G. In contrast to the existing 4G networks, 5G will not target the ubiquitous communication among 7 billion humans, but 5G will focus on the control and steering of 500 billion devices knows as the Internet of Things (IoT). The technical requirements for 5G is not only higher throughput (per user and per area), but the low latency for the control and steering in the range of 1ms is the most critical technical requirement. Furthermore as we want to steer and control e.g. robots and cars, security and resilience become even more important. In order to realize those requirements our following research fields are needed to realize the 5G view we develop here at TU Dresden.
The Tactile Internet enables humans to interact in quasi real-time with cyber-physical systems (CPS) in the real or virtual world over intelligent wide-area communication networks. Such advances go far beyond the current state-of-the-art approaches in computer and engineering sciences: intelligent communication networks and adaptive CPS for quasi real-time co-operations with humans require online mutual learning mechanisms, which are crucial challenges. To tackle these challenges, CeTI will conduct unique interdisciplinary research and will address major open research topics in key areas of the complexity of human control in the human–machine loop, sensor and actuator technologies, software and hardware designs, and the communication networks as the basis for several novel use cases grouped in medicine, industry, and the Internet of Skills.
Our research focuses on studying, designing and evaluating the employment of quantum mechanics in future communication networks. Legacy and future communications present some intrinsic limits because they are based on classical physics. On the other hand, by exploiting entanglement as a communication and computing resource, quantum communication networks can achieve performances that classical communications will never be able to achieve, mainly in terms of security, throughput, latency and energy efficiency.
Since more than a decade we are working on network coding, which is a versatile coding technique for upcoming dynamic transport and cloud architectures. It is a magic juice for wireless mesh, distributed storage, multipath communication, peer-to-peer networks, and many more. Several testbeds and demonstrators have been built to show the benefits of network coding over the old generation of static end to end codes. We are proud to have our own teaching module solely on network coding together with Professor Jorswieck from TUD and hold occasionally tutorials at ICC with Professor Muriel Medard from MIT.
Compressed sensing is the theory that breaks with the Shannon-Nyquist sampling rates and advocates for signal reconstructions from far fewer samples, by solving sparse under-determined systems under certain conditions. Our vision does not emphasize compressed sensing as a signal processing toolkit but rather an intelligent enabler to reduce the massive amount of data transmissions, especially in wireless sensor networks. Sparsity and incoherence are its crux, and this is guaranteed by the temporal and spatial correlations found by default in such networks.
Rising amounts of data sources and traffic require automation in every aspect of the underlying transport network. Machine Learning can answer decision and optimization problems found many fields by making use of patterns in the data flows. Provisioning of networks and deploying functionality within the network can be automated in softwareized networks to adapt to requirements in a timely manner.
Post-Shannon communications advocate for more efficient communication tasks than those proposed by Claude Shannon’s mathematical theory of communication. Within the Shannon paradigm, the communication task is to convey a message from A to B. In some applications of future communication networks, these messages have the goal of producing an outcome, for example, causing a driverless car to break. Post-Shannon communications encompasses all the different communication tasks that allows us to produce the same outcome (i.e. transmit the Gestalt information) more efficiently. Within the Post-Shannon communication, previously useless resources such as common randomness, and noiseless feedback become powerful and capable of increasing the capacity of the channel. Some applications of Post-Shannon communications are identification codes proposed by Ahlswede and Dueck, capable of transmitting and identifying exponentially more messages than message transmission codes; and The Medium is the Message, a proposed technique that makes uses of the channel descriptors of CDMA systems and encodes extra information by choosing a specific pseudo-noise sequence over another one.
Cooperation techniques are a powerful tool to boost the performance in communication networks such as relaying, wireless mesh, virtual antenna fields, etc. We are interested in basic rules of cooperation and adopt those mechanisms to communication networks. Besides the well-known wireless mesh networks, we have carried out research in the field of mobile clouds enabling wireless devices to share physically resources.
In contrast to cloud solutions that are placed at the end of the network architecture, here we investigate agile cloud solutions that are constantly moved at the edge of the network close to the user to facility low latency communication in the range of 1ms for control and steering applications. Our solutions are realized in openstack. Learn more about the Mobile Edge Cloud in our white paper (PDF).
In the future cloud storage and cloud computing will not only be placed in one single place, it will be distributed over several entities, to save energy, to reduce transmission delays, to improve security, and to increase the reliability of a service. In combination with network coding we are interested in developing solutions for a distributed world.
In most of the aforementioned research topics, we collaborate with Massachusetts Institute of Technology (MIT), Aalborg University (AAU), and Budapest University of Technology and Economics (BUTE). Several projects are carried out with our partners from acticom GmbH, Chocolate Cloud ApS, O&O Software, Steinwurf ApS, Meshmerize, and Wandelbots.