Publicação
APPLIED PARALLELIZATION METHODOLOGIES TO GNSS ALGORITHMS STUDY AND IMPLEMENTATION OF DIFFERENT PARALLELIZATION TOOLS FOR ALGORITHM ACCELERATION
| Resumo: | The demand for computational performance continues to grow with the increasing complexity of modern applications. Sequential processing, though conventional, often fails to meet these increasing demands efficiently. This research is motivated by the need to increase computing capabilities of computer systems to overcome these limitations. The study case of a signal modulation algorithm for Global Navigation Satellite System (GNSS) exemplify one scenario where sequential execution results in insufficient processing speeds. Parallelization should offer a viable approach to increase the performance, thus enabling close to real-time processing and improving overall efficiency of the algorithm. Modern hardware architectures, including multi-core Central Processing Unit (CPU)’s and Graphics Processing Unit (GPU)’s, provide ample computational power that can be used to increase computation performance. This study aims to show the adaptation of an existing sequential program for GNSS signal modulation to use the resources available in those components and parallelize the program in order to reduce the time needed for the program’s execution. By accurately identifying performance bottlenecks and applying effective parallelization techniques, the algorithm’s execution time was drastically reduced, achieving speed improvements exceeding 40 times and even 80 times in some cases compared to the program’s baseline version. These optimizations enabled sub-real-time simulation time, marking a significant milestone in the program’s efficiency. With the improved version, it is expected that the algorithm will be turned more useful for current and future GNSS technology development. |
|---|---|
| Autores principais: | Miranda, Daniel Nunes de |
| Assunto: | HPC CUDA Parallel Computing GNSS |
| Ano: | 2025 |
| País: | Portugal |
| Tipo de documento: | dissertação de mestrado |
| Tipo de acesso: | acesso aberto |
| Instituição associada: | Universidade Nova de Lisboa |
| Idioma: | inglês |
| Origem: | Repositório Institucional da UNL |
| Resumo: | The demand for computational performance continues to grow with the increasing complexity of modern applications. Sequential processing, though conventional, often fails to meet these increasing demands efficiently. This research is motivated by the need to increase computing capabilities of computer systems to overcome these limitations. The study case of a signal modulation algorithm for Global Navigation Satellite System (GNSS) exemplify one scenario where sequential execution results in insufficient processing speeds. Parallelization should offer a viable approach to increase the performance, thus enabling close to real-time processing and improving overall efficiency of the algorithm. Modern hardware architectures, including multi-core Central Processing Unit (CPU)’s and Graphics Processing Unit (GPU)’s, provide ample computational power that can be used to increase computation performance. This study aims to show the adaptation of an existing sequential program for GNSS signal modulation to use the resources available in those components and parallelize the program in order to reduce the time needed for the program’s execution. By accurately identifying performance bottlenecks and applying effective parallelization techniques, the algorithm’s execution time was drastically reduced, achieving speed improvements exceeding 40 times and even 80 times in some cases compared to the program’s baseline version. These optimizations enabled sub-real-time simulation time, marking a significant milestone in the program’s efficiency. With the improved version, it is expected that the algorithm will be turned more useful for current and future GNSS technology development. |
|---|
Atividades financiadas
Não foram encontrados projetos financiados.