The \emph{feature interaction problem} arises when two or more independent features interact with each other in an undesirable manner. Feature interactions remain a challenging and important problem in emerging domains of cyber-physical systems (CPS), such as intelligent vehicles, unmanned aerial vehicles (UAVs) and the Internet of Things (IoT), where the outcome of an unexpected interaction may result in a safety failure. Existing approaches to resolving feature interactions rely on priority lists or fixed strategies, but may not be effective in scenarios where none of the competing feature actions are satisfactory with respect to system requirements. This paper proposes a novel \emph{synthesis-based} approach to resolution, where a conflict among features is resolved by \emph{synthesizing} an action that best satisfies the specification of desirable system behaviors in the given environmental context. Unlike existing resolution methods, our approach is capable of producing a desirable system outcome even when none of the conflicting actions are satisfactory. The effectiveness of the proposed approach is demonstrated using a case study involving interactions among safety-critical features in an autonomous drone.

Benjamin Gafford

Grinnell College

United States

Tobias Dürschmid

Carnegie Mellon University, Institute for Software Research

United States

Gabriel A. Moreno

Carnegie Mellon University, USA

Eunsuk Kang

Carnegie Mellon University, USA

United States

Fuzzing or fuzz testing is an established technique that aims to discover unexpected program behavior (e.g. bugs, security vulnerabilities, or crashes) by feeding automatically generated data into a program under test. However, the application of fuzzing to test Model-Driven Software Engineering (MDSE) tools is still limited because of the difficulty of existing fuzzers to provide structured, well-typed inputs, namely models that conform to typing and consistency constraints induced by a given meta-model and underlying modeling framework. By drawing from recent advances on both fuzz testing and automated model generation, we present three different approaches for fuzzing MDSE tools: A graph grammar-based fuzzer and two variants of a coverage-guided mutation-based fuzzer working with different sets of model mutation operators. We have evaluated our fuzzing approaches on a set of real-world MDSE tools. Our experimental results show that all approaches can outperform both standard fuzzers and model generators w.r.t.\ their fuzzing capabilities. Moreover, we found that each of our approaches comes with its own strengths and weaknesses in terms of fault finding capabilities and the ability to cover different aspects of the system under test. Thus the approaches complement each other, forming a fuzzer suite for testing MDSE tools.

Hoang Lam Nguyen

Humboldt-Universität zu Berlin

Nebras Nassar

Philipps-Universität Marburg

Timo Kehrer

Humboldt-Universtität zu Berlin

Germany

Lars Grunske

Humboldt-Universität zu Berlin

Germany

Context: Emergent behaviors are behaviors not included in a system specification, but that can still happen at runtime. When using scenario-based modeling to design a concurrent system, we can detect such behaviors as implied scenarios (ISs). Analogously to emergent behaviors, an IS is not included in the system model but can arise at runtime. If left untreated, ISs can cause damage if they lead to unwanted behaviors, which can, in turn, affect the reliability of the system. Several approaches to detect ISs have been devised. However, existing approaches stop after the detection process and do not go further in the analysis and treatment process. Additionally, they can output several implied scenarios, which can be cumbersome to the user as they are detected and dealt with on a one-on-one basis. Furthermore, since these approaches do not investigate the relationship between different ISs, they could misguide the user on how to deal with such scenarios.

Objective: In this work, we propose a methodology to fill in the literature gap, which is achieved by finding common behaviors (CBs) among detected ISs that lead the system to unexpected behavior. We enable the user to analyze ISs as groups, which allows fixing multiple emergent behaviors at the same time.

Methodology: The methodology consists of the characterization of ISs as families of CBs, comprising three main steps: (i) collect multiple ISs; (ii) detect CBs among them; and, (iii) characterize such CBs as families. Firstly our approach iteratively collects multiple ISs without the need for user interaction. Next, from these collected ISs, we extract the underlying CBs. The CBs are groups of ISs that have common traces, and are defined as shared sequences of messages among various ISs. Following, a characterization process is performed to define families of CBs. For this purpose, we use the Smith-Waterman algorithm, which is used to find which parts of two sequences have the most in common. In our work, the benefits of using this algorithm are two-fold: (i) it can be used to find shared traces of messages among different CBs, and (ii) the score calculated can be used as a clustering metric, which assists the user to define the families of CBs. By these means, we limit the problem space of IS detection by treating the ISs as a group, instead of individually.

Results: We performed seven case studies %with system specifications reported in the literature to validate our methodology, where a total of 1,798 ISs were collected. From these ISs, our methodology was able to identify only 14 families of CBs, where each system specification in those studies had at most three families, and each family required a single fix. Overall, our methodology also managed to significantly reduce the timespan of the detection process from nearly 37 hours to under 24 minutes and mere 3.2s to run our clusterization process in those case studies. Additionally, we provide quantitative evidence that the unwanted behaviors have been effectively removed as all seven case studies had their reliability increase to 100%.

Conclusion: We have proposed a methodology to deal with multiple ISs at large. This is achieved by detecting CBs among these scenarios. Furthermore, we introduce a method to group similar CBs into families, further reducing the elements the user needs to analyze. Thus, our approach allows the user to investigate and treat multiple ISs at once.

Journal Paper: The full paper was published at the Journal of Systems and Software, and is available at https://doi.org/10.1016/j.jss.2019.110425. All tools and data used for the experiments are available at https://git.io/JfCJp.

Link to Publication: https://www.sciencedirect.com/science/article/pii/S0164121219301992

Caio Batista de Melo

University of California, Irvine

André Luiz Fernandes Cançado

University of Brasília

Genaína Nunes Rodrigues

University of Brasília