Fault Injection at Model Level for Signal Data Processing in a Multisensor Instrument for Planetary Exploration

Dependability remains a main concern for the electronic systems involved in aerospace missions or other critical applications where the devices must operate under ionizing radiation or other harsh conditions. While rad-hard and rad-tolerant technologies offer reliable solutions for these tasks, these systems are quite expensive and often have limited performance. Therefore, in recent years there has been a growing trend to use modern devices such as field-programmable gate array (FPGAs) or microprocessors that can meet fault tolerance requirements through hardening techniques and robust designs while offering less expensive and better performing solutions. In this sense, early identification of the weakest elements of the architecture allows designers to apply the necessary hardening techniques to ensure system dependability. In this article, the authors propose a method to assess fault tolerance in digital data processing systems at a high abstraction level. This method is applied to a multisensory system which aims to provide an optimal solution to study meteorological conditions on planetary surfaces. This system, which can read and process signals from meteorological sensors, has an architecture based on a digital lock-in amplifier (DLIA), whose fault tolerance must be studied. The presented method implements a fault injection system in a simulation model of the system under study, emulating the presence of a single-event upset (SEU) in any flip-flop of the circuit. The results of this simulation demonstrate the effects of the faults in the processing system’s output. During the experimental campaigns performed, 561728 faults were injected, of which 47.83% were masked by the data processing architecture. The authors classified the obtained errors and determined which are the most critical elements in the data processing system.