Fault- and attack injection are techniques used to measure dependability attributes of computer systems. An important property of such techniques is their efficiency in exploring the target system's fault- or attack space. As this space is generally very large, pre-injection analysis techniques may be used to effectively explore the space. In this paper, we study two such techniques proposed in the past, namelyinject-on-readandinject-on-write. Furthermore, we propose two new techniques callederror space pruning of signalsanderror space pruning of signals and portsand evaluate their efficiency in reducing the space needed to be explored by injection experiments. These techniques were integrated into MODIFI, a fault- and attack injector targeting Simulink models. To the best of our knowledge, we are the first to evaluate these pre-injection techniques for this kind of injector. The results of our evaluation of 11 Simulink models from the automotive domain and one from the avionics domain, show that the new proposed techniques reduce the fault- and attack space needed to be explored by about 27–49%. Using MODIFI, we then performed injection experiments on two automotive models, as well as an aero engine control model, while elaborating on the results obtained.

Multifunctional neural networks are capable of performing more than one task without changing any network connections. In this paper we explore the performance of a continuous-time, leaky-integrator, and next-generation ‘reservoir computer’ (RC), when trained on tasks which test the limits of multifunctionality. In the first task we train each RC to reconstruct a coexistence of chaotic attractors from different dynamical systems. By moving the data describing these attractors closer together, we find that the extent to which each RC can reconstruct both attractors diminishes as they begin to overlap in state space. In order to provide a greater understanding of this inhibiting effect, in the second task we train each RC to reconstruct a coexistence of two circular orbits which differ only in the direction of rotation. We examine the critical effects that certain parameters can have in each RC to achieve multifunctionality in this extreme case of completely overlapping training data.