Parallel brute-force algorithm for deriving reset sequences from deterministic incomplete finite automata

Abstract

A reset sequence (RS) for a deterministic finite automaton A is an input sequence that brings A to a particular state regardless of the initial state of A . Incomplete finite automata (FA) are strong in modeling reactive systems, but despite their importance, there are no works published for deriving RSs from FA. This paper proposes a massively parallel algorithm to derive short RSs from FA. Experimental results reveal that the proposed parallel algorithm can construct RSs from FA with 16,000,000 states. When multiple GPUs are added to the system the approach can handle larger FA.

Accelerating Finite State Machine-Based Testing Using Reinforcement Learning

“…In the absence of an RS, resetting the SUT may require manual configuration of the SUT or restarting the SUT and, therefore, is considered to be one of the most expensive steps in FSM-based testing [16]. This has led to research regarding algorithms that generate RSs from FSMs [12], [17], [18].…”

“…For example, the existing algorithm for deriving PDSs applies a brute-force search [30]. The state-of-the-art RS generation algorithms require general-purpose graphics processing units and cannot process FSMs having large input domains [12], [17], [18]. Therefore, to enable test case generation from large FSMs, we need scalable methods for deriving RSs and PDSs.…”

“…In this paper, as part of the evaluation, we use such a model: the Engine Status Manager is an embedded program that manages the status of the engine in Océ printers and copiers (a subsidiary of Canon) [11]. This model has thousands of states, and systematically deriving tests with guaranteed fault detection capability from such large complex systems, is a challenge [12], [13], [14]. We address this scalability problem using reinforcement learning.…”

Careful synchronization of partial deterministic finite automata

“…We analyzed these cases using a brute force algorithm known as the successor tree method. See the recent paper by Türker [34] for a description of the method and its modern implementation 4 .…”

Efficient parallel derivation of short distinguishing sequences for nondeterministic finite state machines using MapReduce

“…A variety of studies have recently focused on the use of parallel processing techniques in order to derive CSs in a large-scale context: UIO sequences [17], harmonized state identifiers, and characterizing sets [18], synchronizing sequences [59,60]. For DSs generation, Hierons and Türker [61] and El-Fakih et al [62] introduced independently parallel multithreading implementations of the EA over the Central Processing Unit (CPU) and Graphics Processing Unit (GPU) architectures [63].…”