National and Kapodistrian University of Athens
Σχεδίαση, υλοποίηση και ενσωμάτωση σε σύστημα-σε-μια-ψηφίδα τεχνολογίας FPGA επιταχυντών υλικού υψηλής απόδοσης για συμπίεση υπερφασματικών εικόνων εν πτήσει σε αεροδιαστημικά συστήματα
Abstract
dc:descriptionThe increasing demand for highly detailed information on human and geophysical processes has driven the development of Earth observation satellite hyperspectral sensors. Hyperspectral remote sensing is invaluable for a number of civilian applications including smart agriculture, geology, environmental monitoring, disaster response and recovery. The substantial data volume from high resolution spaceborne hyperspectral imagers compete with the limited on-board storage resources and bandwidth available for the transmission of data to ground stations, making hyperspectral image compression a mission-critical and challenging on-board payload data processing task. At the same time the era of "NewSpace" has emerged in space developments, where launch costs are decreasing and agile approaches are taken in building highly integrated small satellites using low-cost commercial-off-the-shelf parts. Firstly, this thesis demonstrates a novel high data-rate performance hardware accelerator, implementing the CCSDS 123.0-B-1 hyperspectral data compression algorithm as an IP core targeting space-grade FPGAs. By exploiting task-level parallelism of the algorithm, the accelerator implements a reconfigurable fine-grained pipeline in critical feedback loops to achieve high throughput performance in an FPGA optimized RTL design. This established, the state of the art in performance for such single core compressors on space-grade FPGA, achieving 213 Msamples/s (3.3 Gbps@16 bits-per-pixel-per-band) when targeting the Virtex-5QV and 315 Msamples/s (5.04 Gbps@16bpppb) targeting a space-grade Xilinx Kintex UltraScale device, a 5x to 1.5x improvement in performance compared to the previous state-of-the-art.Following work developed for COTS SoC FPGA devices, custom SoC architectures to integrate in parallel the hyperspectral compression accelerator. The proposed architecture is a custom software and hardware co-design and exploits image segmentation to provide increased robustness to data corruption. It enables scalable throughput performance by leveraging segment-level parallelism with multiple compression accelerators in a single chip for improved SWaP-C (Size, Weight, Power and Cost). An implementation is demonstrated on Xilinx Zynq COTS devices, reaching 1387 Msamples/s (22.2Gbps@16bpppb) throughput performance on a Zynq-7045 device with 5 compression cores. This work set the state of the art in performance for CCSDS 123.0-B-1 lossless hyperspectral parallel compression accelerators, outperforming other implementations in equivalent FPGA technology. This is achieved through a detailed study of achievable performance, technology and scalability limits. Finally, another SoC was designed to integrate compression accelerators focused on reliability at the system level, using a soft fault-tolerant LEON processor on a rad-hard Virtex-5 FPGA. The introduced single-chip Data Processing Unit supports self-configuration management and a system-level SEE mitigation strategy is proposed that employs EDAC, TMR and internal scrubbing to guarantee total immunity under extremely harsh radiation environments.
Degree
thesis:*- Grantor dc:publisher
- National and Kapodistrian University of Athens
- Year dc:date
- 2021
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Tsigkanos, Antonis
- Τσίγκανος, Αντώνιος
Subjects
dc:subject × 19- Σχεδίαση υλικού
- Συ-σχεδίαση υλικού-λογισμικού
- Συμπίεση δεδομένων εικόνας εν πτήσει
- Υπερφασματική απεικόνιση
- Επιταχυντές υλικού
- FPGA
- Συστήματα σε μια ψηφίδα
- Σχεδίαση αξιόπιστων συστημάτων
- Hardware design
- Hardware-software co-design
- On-board data compression
- Hyperspectral imaging
- Hardware accelerators
- System on chip
- Embedded systems reliability
- Φυσικές Επιστήμες
- Επιστήμη Ηλεκτρονικών Υπολογιστών και Πληροφορική
- Natural Sciences
- Computer and Information Sciences
Rights
dc:rights- Statement dc:rights
-
- BY_NC_ND
- Language dc:language
- eng
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/48735