Abstract:
Hyperspectral images are widely used in spatial applications and are three-dimensional data structures in which the x and y axes contain spatial information, and the z-ax...Show MoreMetadata
Abstract:
Hyperspectral images are widely used in spatial applications and are three-dimensional data structures in which the x and y axes contain spatial information, and the z-axis contains the spectral bands, which that can reach the order of hundreds of layers. Such images generate a large amount of data, and applying compression algorithms, such as the CCSDS 123 compressor, is highly necessary to deal with the communication and storage constraints of the platforms used to capture hyperspectral images (e.g., a spacecraft). Given that the CCSDS 123 algorithm has a high computational cost, it is necessary to evaluate which processor architectures can deal with it onboard. Given the context above, this work presents an evaluation of performance and power consumption of the CCSDS 123 algorithm running on RISC- V and ARM processors and an evaluation of using two real-time operating systems (FreeRTOS and Zephyr). The experimental results show that, for the development kits used, the RISC- V processor dissipates less power than the ARM processor, which, in turn, offers much higher performance and lower energy consumption than RISC-V. Results also show that FreeRTOS adds a lower overhead to the algorithm execution in comparison to Zephyr when running over the RISC- V processor.
Date of Conference: 24-27 November 2020
Date Added to IEEE Xplore: 08 December 2020
ISBN Information: