Multivariate time series have had many applications in areas from healthcare and finance to meteorology and life sciences. Although deep neural networks have shown excellent predictive performance for time series, they have been criticised for being non-interpretable. Neural Additive Models, however, are known to be fully-interpretable by construction, but may achieve far lower predictive performance than deep networks when applied to time series. This paper introduces FocusLearn, a fully-interpretable modular neural network capable of matching or surpassing the predictive performance of deep networks trained on multivariate time series. In FocusLearn, a recurrent neural network learns the temporal dependencies in the data, while a multi-headed attention layer learns to weight selected features while also suppressing redundant features. Modular neural networks are then trained in parallel and independently, one for each selected feature. This modular approach allows the user to inspect how features influence outcomes in the exact same way as with additive models. Experimental results show that this new approach outperforms additive models in both regression and classification of time series tasks, achieving predictive performance that is comparable to state-of-the-art, non-interpretable deep networks applied to time series.

This paper describes a cloud-based platform that offers evidence-based, personalised interventions powered by Artificial Intelligence to help support efficient remote monitoring and clinician-driven guidance to people over 65 who suffer or are at risk of hearing loss, cardiovascular diseases, cognitive impairments, balance disorders, and mental health issues. This platform has been developed within the SMART-BEAR integrated project to power its large-scale clinical pilots and comprises a standards-based data harmonisation and management layer, a security component, a Big Data Analytics system, a Clinical Decision Support tool, and a dashboard component for efficient data collection across the pilot sites.

Enabled by advances in Cloud technologies, Big Data Analytics Services (BDAS) can improve many processes and identify extra information from previously untapped data sources. As our experience with BDAS and its benefits grows and technology for obtaining even more data improves, BDAS becomes ever more important for many different domains and for our daily lives. Most efforts in improving BDAS technologies have focused on scaling and efficiency issues. However, an equally important property is that of security, especially as we increasingly use public Cloud infrastructures instead of private ones. In this paper we present our approach for strengthening BDAS security by modifying the popular Spark infrastructure so as to monitor at run-time the integrity of data manipulated. In this way, we can ensure that the results obtained by the complex and resource-intensive computations performed on the Cloud are based on correct data and not data that have been tampered with or modified through faults in one of the many and complex subsystems of the overall system.

Despite promoting precise modelling and analysis, architecture description languages (ADLs) have not yet gained the expected momentum. Indeed, practitioners prefer using far less formal languages like UML, thus hindering formal verification of models. One of the main issues with ADLs derives from process algebras which practitioners view as having a steep learning curve. In this paper, we introduce a new ADL called XCD which enables designers to model their software architectures through a Design-by-Contract approach, as for example in the Java Modelling Language (JML). We illustrate how XCD can be used in architectural modelling and analysis using the Aegis combat software system.

Despite being a widely-used language for specifying software systems, UML remains less than ideal for software architectures. Architecture description languages (ADLs) were developed to provide more comprehensive support. However, so far the application of ADLs in practice has been impeded by at least one of the following problems: (i) advanced formal notations, (ii) lack of support for complex connectors, and (iii) potentially unrealizable designs. In this paper we propose a new ADL that is based on Design-by-Contract (DbC) for specifying software architectures. While DbC promotes a formal and precise way of specifying system behaviours, it is more familiar to practising developers, thus allowing for a more comfortable way of specifying architectures than using process algebras. Furthermore, by granting connectors a first-class status, our ADL allows designers to specify not only simple interaction mechanisms as connectors but also complex interaction protocols. Finally, in order to ensure that architectural designs are always realizable we eliminate potentially unrealizable constructs in connector specifications (the connector “glue”).

Design-by-Contract (DbC) gained wide familiarity among software developers for specifying software. It aids in documenting the behaviour of class methods as contracts between clients of the methods (pre-) and their suppliers (post-condition). This not only allows developers to document software behaviour precisely at such a high-level that can more easily be communicated, but also enables the formal verification of the behaviour. In this paper, we provide a comprehensive extension to DbC so that it can also be applied to the level of software architecture design. We illustrate this through our architecture description language XCD. Components in XCD have four different types of interfaces: provided and required interfaces of methods or emitter and consumer interfaces of events where methods/events are contractually specified. Contract specification is separated into functional and interaction contracts thus modularising the functional and interaction component behaviours. Furthermore, treating interaction protocols as connectors, XCD allows to specify connectors with interaction contracts that participating components adhere to. The formal semantics of XCD are defined using Finite State Process (FSP) thus enabling formal analysis of contractually specified software architectures for quality properties, e.g., deadlock.

Future deployments of wireless sensor network (WSN) infrastructures for environmental or event monitoring are expected to be equipped with energy harvesters (e.g. piezoelectric, thermal, photovoltaic) in order to substantially increase their autonomy. In this paper we derive conditions for energy neutrality, i.e. perpetual energy autonomy per sensor node, by balancing the node's expected energy consumption with its expected energy harvesting capability. Our analysis assumes a uniformly-formed WSN, i.e. a network comprising identical transmitter sensor nodes and identical receiver/relay sensor nodes with a balanced cluster-tree topology. The proposed framework is parametric to: (i) the duty cycle for the network activation; (ii) the number of nodes in the same tier of the cluster-tree topology; (iii) the consumption rate of the receiver node(s) that collect (and possibly relay) data along with their own; (iv) the marginal probability density function (PDF) characterizing the data transmission rate per node; (v) the expected amount of energy harvested by each node. Based on our analysis, we obtain the number of nodes leading to the minimum energy harvestingrequirement for each tier of the WSN cluster-tree topology. We also derive closed-form expressions for the difference in the minimum energy harvesting requirements between four transmission rate PDFs in function of the WSN parameters. Our analytic results are validated via experiments using TelosB sensor nodes and an energy measurement testbed. Our framework is useful for feasibility studies on energy harvesting technologies in WSNs and for optimizing the operational settings of hierarchical WSN-based monitoring infrastructures prior to time-consuming testing and deployment within the application environment.

This paper describes the potential impact of the Internet of Things (IoT) technologies and architecture on factory automation. Whereas, IoT use cases range from intelligent infrastructure and smart cities to health care and shopping assistants, it is important to note that factory automation could benefit as well from an IoT approach. In this paper, we argue that there will not be one IoT but many IoTs that could differ in the type of infrastructure they are running or applications they support. In IoT@Work we focus on the potential of making manufacturing environments more agile and flexible. We explain how the IoT-centric architecture for manufacturing also needs a deep understanding of the manufacturing system and its state today. We, therefore, do a reverse engineering based on the requirements and the description of the agility expected in the automation system itself.

We present JUPITER, a tool for analysing multi-constrained systems. JUPITER was built to explore three basic ideas. First, how to use controller synthesis so as to find the exact conditions under which a particular constraint will be satisfied. Second, how to successively refine the models used for the controller synthesis so as to obtain a series of more easily understandable and more robust controllers. Last but not least, how to structure & explain the synthesised controllers and provide hints to designers for further optimisations through the use of machine learning techniques. Thus, JUPITER can help in the design and analysis of multi-constraint systems through the automatic synthesis of control logic for certain of the constraints and the aid it provides to designers for discovering further optimisations. The controllers it synthesises can be easily implemented on top of a standard real-time OS.

We present a new scheduler architecture, which permits adding QoS (quality of service) policies to the scheduling decisions. We also present a new scheduling synthesis method which allows a designer to obtain a safe scheduler for a particular application. Our scheduler architecture and scheduler synthesis method can be used for heterogeneous applications where the tasks communicate through various synchronization primitives. We present a prototype implementation of this scheduler architecture and related mechanisms on top of an open-source OS (operating system) for embedded systems.

When designing complex software systems that provide multiple non-functional properties, it is usual to try to reuse (and finally compose) simpler existing designs, which deal with each of these properties in solitude. The paper describes a method for automatically and quickly identifying all the different ways one can compose such designs, with the aid of a model checker.

A method is presented for the automatic construction of all possible valid compositions of different middleware software architectures. This allows reusing the latter in order to create systems providing a set of different non-functional properties. These compositions are constructed by using only the structural information of the architectures, i.e. their configurations. Yet, they provide a valuable insight on the different properties of the class of systems that can be constructed when a particular set of non-functional properties is required.