5GEnabled Real Time Communications for "Tactile Internet" (5G-Remote)

• Applicants: Professor Dr.-Ing. Eckhard Grass; Professor Dr.-Ing. Rolf Kraemer
• Subject Area: Communication Technology and Networks, High-Frequency Technology and Photonic Systems, Signal Processing and Machine Learning for Information Technology
• Term: since 2021
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 457407152

The introduction of the 5th generation of mobile communications systems (5G) will pose a crucial challenge for real-time communications. The most important requirement for many of these real-time applications is a guaranteed minimum network round trip delay (RTD) of ≤ 1ms. In order to achieve the required RTD, the resources for the end-to-end (E2E) network must be jointly optimized. We argue that URLLC, as proposed for 5G, is not sufficient to meet the requirements for tactile Internet applications (TI). In this project, flexible and programmable network architectures with efficient real-time Virtualized Radio Resource Management (vRRM) techniques are investigated, simulated and prototypically implemented to achieve the desired performance parameters for TI services. AI frameworks and libraries are also used to achieve improved virtualization by using network data. However, we will not conduct our own AI research. The optimization of the E2E wireless network connection requires the investigation of all network resources, especially communication, computation and memory resources, which means a huge amount of work. Therefore, we will focus on the investigation and development of a reliable hypervisor and real-time planning of network resources for the cMTC (Critical Machine Type Communication) in order to achieve necessary improvements over URLLC (Ultra-Reliable Low-Latency Communication) service quality. This requires moving critical parts of the cloud control application to the edge server near the CPS. We do not intend to investigate the specific mobility of the end user, so we can assume a static or nomadic relationship between the end user and the Edge Server. The project will investigate vRRM techniques and algorithms to solve the necessary network RTD optimization problem for TI-specific real-time wireless communication in a factory automation use case. The original optimization problem is formulated as a Markov Decision Process (MDP)-based constrained coverage optimization problem in which the optimal solution is sought. In order to prove the usability of AI-based approaches, online learning algorithms are applied that approach the optimal solution with low computational complexity. With the help of open source software like OpenStack, OPNFV and OpenFog, optimal and near-optimal solutions are anticipated. For the investigation of AI support for hypervisor planning policies and the use of learning algorithms we will use AI frameworks like TensorFlow or CNTK for experimental DNN modeling (Deep Neural Networks) and AI libraries like Torch or MLPack for prototypical implementations and evaluations.

• DFG Programme: Research Grants