Program for Project-Related Personal Exchange (PPP)
This page lists the running project funded by German Academic Exchange Service (DAAD).
Computation Offloading Designs for Autonomous Drone Swarms
| Project Period | 01.01.2021-12.31.2022 |
| Principal Investigator in Germany | Prof. Dr. Jian-Jia Chen |
| Principal Investigator abroad | Prof. Guthemberg Silvestre |
| Institution abroad | Ecole Nationale de l'Aviation Civile (ENAC) |
| Project Members in Germany | Dr.-Ing. Kuan-Hsun Chen, Niklas Ueter, Junjie Shi, Abdur Rehman |
Project Description
This joint project aims to propose novel techniques for emerging services on swarms of autonomous drones while considering the resource constraints. We intend to focus our efforts to design, evaluate and integrate suitable computation offloading techniques on top of edge resources of drones. In order to pursue these challenging goals, we should conduct interdisciplinary, collaborative research, which includes open research problems in mobile computing, cyber-physical systems, multi-objective optimization, dynamic schedule and adaptive edge strategies.
ACCo-DT: Application and Compilation Co-design for Decision Trees
| Project Period | 01.01.2021-12.31.2022 |
| Principal Investigator in Germany | Dr.-Ing. Kuan-Hsun Chen |
| Principal Investigator abroad | Prof. Jenq-Kuen Lee |
| Institution abroad | National Tsing Hua University (NTHU) |
| Project Members in Germany | Christian Hakert, Prof. Dr. Jian-Jia Chen |
Project Description
This joint project aims to provides software solutions to study the interplay of application and compilation and form a joint research project to explore the opportunities of joint optimization methods for Decision Trees and relevant machine learning applications.
Since effect of different synthesis methods is eventually carried out by the underlying compilation. Any careless optimization on the application layer may be mitigated or even masked by the inherent compilation principles. In contrast, an ad-hoc compilation optimization may not be affected if the execution pattern of applications can not be recognized.