Projecte matriculat
Títol: Design and implementation using AI Engine over a Versal AI Edge SoC of a multi-beam Time-Domain Digital Beamformed for 5G
Director/a: BERENGUER I SAU, JORDI
Departament: TSC
Títol: Design and implementation using AI Engine over a Versal AI Edge SoC of a multi-beam Time-Domain Digital Beamformed for 5G
Data inici oferta: 18-09-2025 Data finalització oferta: 18-04-2026
Estudis d'assignació del projecte:
GR ENG SIS TELECOMUN
| Tipus: Individual | |
| Lloc de realització: Fora UPC | |
| Supervisor/a extern: Alessandro Modigliana | |
| Institució/Empresa: SATELIOT | |
| Titulació del Director/a: Telecommunications engineering master | |
| Paraules clau: | |
| 5G, Array, Antennas, Beamforming | |
| Descripció del contingut i pla d'activitats: | |
| In this project it will first be studied the theory and then characterised the behaviour and performance of a multi-beam time-domain digital beamformer for an array antenna using Matlab. Then it will be explored and compared with the various algorithm implementations of a time-domain digital beamforming algorithm, and the selected one will be implemented in Matlab. As a final point the selected algorithm will be adapted to a Digital Signal Processor, adapting the algorithm to take advantage of vector/parallel processing in order to have a real-time and low power implementation suitable for a deployment in a commercial environment.
The project will be structured in the following parts: 1. Study of the fundamental beam forming theory 2. Characterisation of beam patterns for different beam steering, for single and multiple beams, and comparison with provided CST simulations 3. Characterisation of the beam-to-beam interference in a multi-beam scenario 4. Implement techniques for beam-to-beam interference reduction 5. Compare different digital beam forming algorithm implementations and implements the selected one in Matlab using vector instructions and external 5G test vectors 6. Rewrite the digital beam forming algorithm in C++ and update the algorithm to take advantage of vector instructions supported by the AI Engine nodes. 7. Deploy the software on a Versal AI Edge SoC and compare the antenna patterns against the Matlab version using the same 5G test vectors (in a single beam and multi beam scenario) 8. Profiling and further code optimization on Versal AI Edge SoC 9. Emulate Real-Time input (Az, El) and perform beamforming update at different periodicity for comparison. |
|
| Orientació a l'estudiant: | |
| Horari d'atenció a estudiants per a l'assignació de projecte: |
|