Integration of COTS Processing Architectures in Small Satellites for Onboard Computing Using Fault Injection Testing Methodology

Abstract

AbstractOver the last decade, the interest of the space industry has increased towards smaller missions with reduced instruments. Advances in miniaturization technologies for electronics have increased the development of small spacecrafts, such as nanosatellites and microsatellites, from commercial-off-the-shelf (COTS) devices of low size, weight, and power. However, the effects of harsh space environment conditions on COTS electronic devices limit their use in high-performance multicore and heterogenous architectures, such as onboard edge computing based on Graphics Processing Units (GPU) for Artificial Intelligence (AI) applications. This article analyses the main considerations for adopting the Fault Injection Testing (FIT) methodology for software-intensive developments based on COTS, primarily intended to test embedded faults that emulate potential faults triggering to allow verification of fault detection, isolation, reconfiguration, and recovery capabilities in fault-tolerant and safety-critical systems. Two variants of techniques can be considered in the FIT methodology, based on hardware and software, but it is the software FIT variant (SFIT) that provides an inexpensive and time-efficient framework to replicate the fault triggering effects by injecting faults into code, data, and interfaces. A proposal is presented to apply the FIT methodology in the testing of future versions of the UPMSat-2 microsatellite payload at Spanish University Institute of Microgravity “Ignacio Da Riva” (IDR/UPM). The scope of the proposed methodology aims to facilitate the future validation of satellite real-time control systems, designed with model-driven engineering processes, and the future onboard integration of high-performance COTS GPUs to process sensor-fusion payload data and to implement auxiliary controllers with AI computing techniques.KeywordsFault Injection Testing (FIT)Commercial-off-the-self (COTS)Small satelliteOnboard computingGraphics Processing Unit (GPU)Fault-tolerantSafety-criticalReal-time control systemSensor-fusion payload data