Estudio de la personalización del RISC-V para aplicaciones espaciales
Loading...
Official URL
Full text at PDC
Publication date
2024
Advisors (or tutors)
Editors
Journal Title
Journal ISSN
Volume Title
Publisher
Citation
Abstract
Durante los últimos años, y debido al avance de la tecnología, se ha producido un aumento en el uso de pequeños satélites para la teledetección, mediante técnicas de adquisición de datos de la superficie terrestre, con el fin de prevenir y actuar frente a diversos problemas climáticos y ambientales.
Una gran variedad de estos pequeños satélites, como los CubeSat, cuentan con diversas restricciones en cuanto a tamaño, peso, energía y tiempo de procesamiento, lo que hace crucial la optimización de estos factores.
Este Trabajo de Fin de Grado se ha centrado en esta optimización del tiempo de procesamiento de la plataforma, utilizando para ello un procesador basado en la arquitectura RISC-V. Esta cuenta con numerosas ventajas frente a otras de carácter privativo por las posibilidades de personalización y la capacidad de entender en profundidad su funcionamiento.
Haciendo uso de estas características, hemos podido optimizar dos algoritmos de detección de nubes, utilizados para reducir la sobrecarga en el ancho de banda de comunicación del satélite al realizar la transmisión de los datos con la estación en tierra. Estas mejoras han consistido en una optimización a nivel software de uno de los algoritmos, y una mejora a nivel hardware del otro algoritmo, con la implementación de instrucciones no estándar, haciendo así uso de las posibilidades de customización que ofrece esta arquitectura.
Tras la realización del proyecto se ha conseguido una mejora significativa en el tiempo de ejecución de ambos algoritmos, subrayando el potencial de los procesadores RISC-V en este tipo de aplicaciones, gracias a su versatilidad y carácter abierto.
In recent years, due to technological advancements, there has been an increase in the use of small satellites for remote sensing, employing techniques to acquire data from the Earth’s surface to prevent and address various climatic and environmental issues. A wide variety of these small satellites, such as CubeSats, have several constraints in terms of size, weight, power, and processing time, making the optimization of these factors crucial. This Final Degree Project has focused on optimizing the processing time of the platform using a processor based on the RISC-V architecture. This architecture offers numerous advantages over proprietary ones due to its customization possibilities and the ability to deeply understand its functioning. Utilizing these characteristics, we were able to optimize two cloud detection algorithms used to reduce the communication bandwidth overload of the satellite when transmitting data to the ground station. These improvements consisted of a software-level optimization of one algorithm and a hardware-level enhancement of the other algorithm, with the implementation of non-standard instructions, thus leveraging the customization possibilities offered by this architecture. After completing the project, a significant improvement in the execution time of both algorithms was achieved, highlighting the potential of RISC-V processors in this type of application, thanks to their versatility and open source nature.
In recent years, due to technological advancements, there has been an increase in the use of small satellites for remote sensing, employing techniques to acquire data from the Earth’s surface to prevent and address various climatic and environmental issues. A wide variety of these small satellites, such as CubeSats, have several constraints in terms of size, weight, power, and processing time, making the optimization of these factors crucial. This Final Degree Project has focused on optimizing the processing time of the platform using a processor based on the RISC-V architecture. This architecture offers numerous advantages over proprietary ones due to its customization possibilities and the ability to deeply understand its functioning. Utilizing these characteristics, we were able to optimize two cloud detection algorithms used to reduce the communication bandwidth overload of the satellite when transmitting data to the ground station. These improvements consisted of a software-level optimization of one algorithm and a hardware-level enhancement of the other algorithm, with the implementation of non-standard instructions, thus leveraging the customization possibilities offered by this architecture. After completing the project, a significant improvement in the execution time of both algorithms was achieved, highlighting the potential of RISC-V processors in this type of application, thanks to their versatility and open source nature.
Description
Trabajo de Fin de Grado en Ingeniería de Computadores e Ingeniería Informática, Facultad de Informática UCM, Departamento de Arquitectura de Computadores y Automática, Curso 2023/2024