A fault-tolerant CCSDS 123 hardware accelerator for space applications

Abstract

Remote sensing techniques in space applications utilize HSIs (Hyperspectral Images) to gather vast amounts of Earth data. The high data volumes associated with HSIs pose significant challenges for storage and processing capabilities in space systems, emphasizing the need for efficient compression. Given the susceptibility of space-based systems to faults due to harsh environmental conditions, fault tolerance mechanisms are essential for ensuring reliable operation. This work presents a low-cost and fault-tolerant CCSDS 123 HSI compressor that employs TMR (Triple Modular Redundancy) and Hamming ECC (Error Correcting Code) to mitigate SEUs (Single Event Upsets). We present the compressor in a standard version, followed by partially protected and fully protected versions, each incorporating varying hardening levels in different circuit components. We performed a fault injection campaign to assess the reliability of all implementations. Results show that the standard version exhibited a high error rate of 97.9%, which presented a significant reduction in the partially hardened versions, reaching no error propagation in the fully hardened version. The standard solution presented the lowest resource utilization and can process 20.57 MSa/s, considering the Hyperion image configuration. Furthermore, all implementations accelerated the application, resulting in a performance improvement of up to 24$\times$ compared to a software solution.