Thesis Code: 20018
Thesis Type: Master Thesis for Telecommunication/Electronic Engineering, Computer Science or equivalent
Research Area: Advanced Computing, Photonics and Electromagnetics (CPE)
- MS students in Telecommunication Engineering, Electronic Engineering, Computer Science or equivalent
- Experience with main programming languages (Python/Matlab /Fortran/C/C++)
- Experience in programming GPU and or FPGA (VHDL, CUDA, OpenCL, etc.)
- Basic knowledge of electromagnetism
Computational electromagnetics (CEM) is the base of the design of all modern telecommunications systems and, in general, of electromagnetic applications. Increasingly sophisticated and fast algorithm solving Maxwell’s equations are needed to develop innovative technologies and solutions. Traditional acceleration strategies for CEM involve distributed computing methods (such as MPI) and shared memory programming paradigms (e.g. OpenMP) in multi-threaded/multi-core or even HPC hardware. Additional improvements have also been established with graphic processing units (GPUs). The aim of this project is to implement a parallelised computational electromagnetics (CEM) solver, for well-known techniques, such for example the Method-of-Moments (MoM), using hardware acceleration strategies. For that purpose, properly selected CEM algorithms must be ported, implemented and run in hardware and efficiently integrated in the computational environment. These acceleration techniques are focussed on applying devices such as FPGAs and GPUs to improve the memory and run-times associated with conventional solvers.
This thesis aims to demonstrate the hardware acceleration of best candidate CEM algorithms to achieve higher global performances
- Denonno et al., “GPU-based acceleration of computational electromagnetics codes”, International Journal of Numerical Modelling: Electronic Networks, Devices and Fields, 2013; 26:309–323
Contact: send a resume with attached the list of exams to firstname.lastname@example.org specifying the thesis code and title.