Monte-carlo techniques for falsification of temporal properties of non-linear hybrid systems

Abstract

We present a Monte-Carlo optimization technique for finding inputs to a system that falsify a given Metric Temporal Logic (MTL) property. Our approach performs a random walk over the space of inputs guided by a robustness metric defined by the MTL property. Robustness can be used to guide our search for a falsifying trajectory by exploring trajectories with smaller robustness values. We show that the notion of robustness can be generalized to consider hybrid system trajectories. The resulting testing framework can be applied to non-linear hybrid systems with external inputs. We show through numerous experiments on complex systems that using our framework can help automatically falsify properties with more consistency as compared to other means such as uniform sampling.

Querying Parametric Temporal Logic Properties on Embedded Systems

“…Similarly, a lot of research has been invested on testing methods for Model Based Development (MBD) of embedded systems [3]. However, the temporal logic testing of embedded and hybrid systems has not received much attention [20,21,4,22]. Parametric temporal logics were first defined over traces of finite state machines [23].…”

“…In [4], we investigated a new approach for testing embedded and hybrid systems against formal requirements in Metric Temporal Logic (MTL) [5]. Our work was premised on the need to express complex design requirements in a formal logic for both requirements analysis and requirements verification.…”

“…Thus, only stochastic optimization techniques can be employed for solving the optimization problem and, in turn, for solving the initial falsification problem. In our previous research [7,8,4], we have explored the applicability of various stochastic optimization methods to the MTL falsification problem with great success.…”

“…In order to solve the resulting optimization problem, we utilize our MTL falsification toolbox S-TaLiRo [10], which contains a number of stochastic optimization methods [7,8,4]. Finally, we demonstrate our framework on a challenge problem from the industry [11] and we present some experimental results on a small number of benchmark problems.…”

Simulating Insulin Infusion Pump Risks by In-Silico Modeling of the Insulin-Glucose Regulatory System

“…Robustness-Guided Model Checking: We use robustness-guided sampling assuming that faults are non-deterministic [28,4] to explore the possible worst case scenarios involving a combination of faults. Robustness guided sampling is based on the idea of providing real-valued robustness semantics to formulas in metric temporal logic [30,16,17].…”