After successful completion of the course, students are able to select suitable optimization methods in general. Optimization and analysis is also the core of successful SDN-Twin networking and is completely based on well-established methods and algorithms provided by the world-wide scientific community. No source can be richer to deliver the right solution to obtain all required tailored properties that are demanded. The list of scientific disciplines is increasing fast. Lifetime design and operation make the difference. Networking is everything that can be described by relationships that finally becomes a graph topology. With SDN-Twin every network design becomes the operating life-time stable network. SDN-Twin networking as well as digital twin optimization is based on mathematics, operations research and metaheuristics based on the basic science disciplines, nature and society.
This course will be of very special interest because additionally to real world physical network applications, SDN-Twin can be used as physical accelerator network for logical or strategic problems. In this context mathematical-oriented methods can be accelerated by using the SDN-Twin networking constellations CPU/GPU/DPU (central-universal, graphic-parallel, data-centric). This acceleration is possible because many matrix calculations can be done in parallel. In linear programming for instance, the CPU selects a new pivot point in an n-dimensional space and then one or more GPUs perform the bulk work in record time. A DPU would additionally be helpful to distribute GPUs over a graph structure. That is one point-of-interest; an other one is the design of digital application or service twins where metaheuristic optimization models apply often much better.
Optimization based on mathematics, operations research, the basic science discplines, nature and society.
Lecture with discussion and joined leaning from new insights in the small SDN-Twin semester projects assigned to students that want to take an exam.
Slides and additional information
Software-defined-networking with digital twins is a complete new way of networking for many kinds of networks. Teaching this new method can only be effective when at least all major issues are addressed. There must be a fast and steep learning curve. This new topic is covered in a set of optional lectures for which one can collect additional credits. In the Master and PhD programs, there is not a need to collect so many credits. In fact, this is a complete special topic study embedded in a collection of voluntary lectures. In order that attendees interested to be trained in all aspects of SDN-Twin networks can profit, an appropriate mode of examination will be introduced. The examination will be performed as special examination events with additional audience in lecture room EI 7 during lecture-free days after and before a semester. In addition, also such events in smaller rooms will be organized on request. Both in WS and SS there are SDN-Twin lectures. On the SDN-Twin homepage, additional information will be available to prepare for the examination questions. In each semester, candidates can take an exam for up to 4 visited lectures with an assigned topic consisting of a written report (about 20 pages), a presentation (20 minutes) and answering questions on the content of the lectures indicated in the exam registration (10 minutes). Each visited optional lecture will be honoured by 3 ECTS, so that also all surplus optional lectures will appear as certificate for the special SDN-Twin education. Of course listeners are welcome to attend all examination events.
Not necessary
This is a new way of IP and Internet-of-things networking, since SDN-Twin is purely based on CPU/GPU/DPU processing with memory hierarchy and transmission technologies. Therefore, current networking knowledge is helpful, but not much relevant. With respect to informatics courses, SDN-Twin will be an important case to be aware of, because several topics of SDN-Twin are here already well-known..