Quality of Service analysis of composed software systems is an active research area, with the goal of evaluating and improving performance and resource allocation in serviceoriented applications, namely, in the glue code –coordination layer– of such systems. Stochastic Reo offers constructs for
service coordination and allows the specification of stochastic values for channels. But its state-of-the-art semantic models fail in several (important) ways. In this paper, we will see how Interactive Markov chains (IMC), proposed as astochastic compositional model of concurrency, can be effectively used to serve as a compositional semantic model for Stochastic Reo. Treating IMC as a direct semantic model, gives rise to more faithful models and has obvious efficiency advantages. Moreover, tool support that exists for IMC is made available, without significant effort, to verify and reason about the coordination layer modelled as Reo connectors.

Bidirectional transformation (BX) is a very active area of research interest. There is not only a growing body of theory, but also a rich set of tools supporting BX. The problem now arises that there is no commonly agreed-upon suite of tests or benchmarks that shows either the conformance of tools to theory, or the performance of tools in particular BX scenarios. This paper sets out to improve the state of affairs in this respect, by proposing a template and a set of required criteria for benchmark descriptions, as well as guidelines for the artifacts that should be provided for each included test. As a proof of concept, the paper additionally provides a detailed description of one concrete benchmark.

Diagnosing unwanted behaviour in Multi-Agent Systems (MASs) is a crucial task to assure the correct operation of a system. A light-weight technique inspired by the software-engineeringoriented techniques, the we have coined Extended Spectrum-based Fault Localisation for Multi-Agent Systems (ESFL-MAS) can be used to shorten the diagnose cycle by reducing the testing effort. As the technique relies on minimal information about the system, its diagnostic accuracy is inherently limited. In this paper, we study the impact of different similarity coefficients that are applied to the system execution spectra as we try to assess the correct operation of an agentbased application. The studied coefficients are proposed in the literature and we study the impact of them on the accuracy of spectrum-based fault localisation applied to multi-agent systems. Our experimental evaluation shows that three out of 42 (Jaccard, Ochiai, and Sorensen-Dice) have the most effective and stable performance throughout the variation of the number of passed and failed events.

Adapting technological environments to users is a concern since Mark Weiser launched the concept of ubiquitous computing and, in order to do that, is necessary to understand users' characteristics. In this context, the purpose of this paper is to present a study about students' mobility habits within a university campus, having the intention of getting insights towards the best place to set an interactive public display and of predicting the main characteristics of the audience that will be present on that spot in forthcoming periods. Thus, the envisioned results of this work will allow the adaptation of the contents exhibited on the device to the audience. To perform the study, a set of logs of accesses to the university's Wi-Fi was used, data mining techniques were implemented and forecasting models were built, using the line of work suggested by the CRISP-DM methodology. As result, students profile were built based on past wireless accesses and on their scholar schedules, and three time series models were used (Holt-Winters, Seasonal Naive and Simple Exponential Smoothing) to predict the presence of students on the envisioned spot in future periods.

We present a formal verification approach for detecting design issues related to user interaction, with a focus on user interface of medical devices. The approach makes a novel use of configuration diagrams proposed by Rushby to formally verify important human factors properties of user interface implementation. In particular, it first translates the software implementation of user interface into an equivalent formal specification, from which a behavioral model is constructed using theorem proving; human factors properties are then verified against the behavioral model; lastly, a comprehensive set of test inputs are produced by exploring the behavioral model, which can be used to challenge the real interface implementation and to ensure that the issues detected in the behavior model do apply to the implementation. We have prototyped the approach based on the PVS proof system, and applied it to analyze the user interface of a real medical device. The analysis detected several interaction design issues in the device, which may potentially lead to severe consequences.

This paper presents a generic infusion pump user interface (GIP-UI) architecture that intends to capture the common characteristics and functionalities of interactive software incorporated in broad classes of infusion pumps. It is designed to facilitate the identification of use hazards and their causes in infusion pump designs. This architecture constitutes our first e!ort at establishing a model-based risk analysis methodology that helps manufacturers identify and mitigate use hazards in their products at early stages of the development life-cycle. The applicability of the GIP-UI architecture has been confirmed in a hazard analysis focusing on the number entry software of existing infusion pumps, in which the GIP-UI architecture is used to identify a substantial set of user interface design errors that may contribute to use hazards found in infusion pump incidents.