Projects per year
In phase I of project C1 the investigations have been considering moving transceivers without taking into account human blockage, whereas the investigations on human blockage have assumed static transceivers. These two scenarios will be combined into a scenario with human blockage when one or both transceivers are on the move. The corresponding device discovery and beam searching algorithms considered in the second phase of C1 may exploit predictions of the movements for which Reinforcement Learning algorithms can be used. We will build on the results on human blockage investigated in project B1. The project will focus on three tasks taking into account the findings from phase I. In the first task we will enhance the device discovery and beam searching algorithms by applying Reinforcement Learning. In the second we will investigate software and hardware optimization and acceleration of the collected algorithms developed for device discovery based on algorithm-architecture co-design, which will require close cooperation of both PIs participating in this project. Based on these results we will explore ASIC concepts for Deep Learning inference accelerators in a third task.
|Effective start/end date||01.01.19 → …|
DFG Research Classification Scheme
- 408-02 Communications, High-Frequency and Network Technology, Theoretical Electrical Engineering