The Internet of Things is one of the core drivers of variability modelling and requires explicit mechanisms to manage it. A key technology for addressing this variability is product line engineering. This approach uses a reference architecture to establish a well-designed set of assets that fit together, the Software Product Line (SPL). One of the limitations of variability models is they do not provide information about the quality of new products or how they achieve stakeholder requirements. Several approaches tackle this issue by integrating variability models with goal models. The main challenge is conciliating the different variability perspectives to make the joint use of both models possible without the loss of information or alterations to the models' semantics. In this work, we present a framework for analysing and optimising SPL products considering stakeholders' requirements that respects the semantics of both models. The framework is based on Integer Linear Programming (ILP), a field of mathematical programming. Variability and goal models are formalised as a set of linear constraints and are linked using mapping functions. As a proof of concept, we present a tool that takes both models and mapping functions to generate an ILP problem that can be solved using Matlab.

Edge Computing proposes to use the nearby devices in the frontier/Edge of the access network for deploying application tasks of IoT-based systems. However, the functionality of such cyber–physical systems, which is usually distributed in several devices and computers, imposes specific requirements on the infrastructure to run properly. The evolution of an application to meet new user requirements and the high diversity of hardware and software technologies in the IoT/Edge/Cloud can complicate the deployment of continuously evolving applications. The aim of our approach is to apply Multi Layer Feature Models, which capture the variability of applications and the software and hardware infrastructure, to support the deployment in edge-based environments of cyber–physical applications. With this multi-layered approach is possible to support the evolution of application and infrastructure independently. Considering that IoT/Edge/Cloud infrastructures are usually shared by many applications, the deployment process has to assure that there will be enough resources for all of them, informing developers about the feasible alternatives. We provide four modules so that the developer can calculate what is the configuration of minimal set of devices supporting application requirements of the evolved application. In addition, the developer can find what is the application configuration that can be hosted in the current infrastructure. The successive solutions of continuous deployment generated by our approach pursue the reduction of the system energy footprint and/or execution latency.

Industrial Augmented Reality (IAR) is a key enabling technology for Industry 4.0. However, its adoption poses several challenges because it requires the execution of computing-intensive tasks in devices with poor computational resources, which contributes to a faster draining of the device batteries. Proactive self-adaptation techniques could overcome these problems that affect the quality of experience by optimizing computational resources and minimizing user disturbance. In this work, we propose to apply ProDSPL, a proactive Dynamic Software Product Line, for the self-adaptation of IAR applications to satisfy the quality requirements. PRODSPL is compared against MODAGAME, a multi-objective DSPL approach that uses a genetic algorithm to generate quasi-optimal feature model configurations at runtime. The evaluation with randomly generated feature models running on mobile devices shows that PRODSPL gives results closer to the Pareto optimal than MODAGAME.

In order to lower latency and reduce energy consumption, edge computing proposes offloading some computation-intensive tasks usually performed in the cloud onto nearby devices in the frontier/edge of the access networks. However, the current task offloading approaches are often quite simple. They neither consider the high diversity of hardware and software technologies present in edge network devices nor take into account that some tasks may require some specific software and hardware infrastructure to be executed. This article proposes a task offloading process that leans on software product line technologies, which are a very good option to model the variability of software and hardware present in edge environments. First, our approach automates the separation of application tasks, considering the data and operation needs and restrictions among them, and identifying the hardware and software resources required by each task. Second, our approach models and manages separately the infrastructure available for task offloading, as a set of nodes that provide certain hardware and software resources. This separation allows to reason about alternative offloading of tasks with different hardware and software resource requirements, in heterogeneous nodes and minimizing energy consumption. In addition, the offloading process considers alternative implementations of tasks to choose the one that best fits the hardware and software characteristics of the available edge network infrastructure. The experimental results show that our approach reduces the energy consumption in the user node by approximately 41%–62%, and the energy consumption of the devices involved in a task offloading solution by 34%–48%.

Advanced smart home appliances and new models of energy tariffs imposed by energy providers pose new challenges in the automation of home energy management. Users need some assistant tool that helps them to make complex decisions with different goals, depending on the current situation. Multi-agent systems have proved to be a suitable technology to develop self-management systems, able to take the most adequate decision under different context-dependent situations, like the home energy management. The heterogeneity of home appliances and also the changes in the energy policies of providers introduce the necessity of explicitly modeling this variability. But, multi-agent systems lack of mechanisms to effectively deal with the different degrees of variability required by these kinds of systems. Software Product Line technologies, including variability models, has been successfully applied to different domains to explicitly model any kind of variability. We have defined a software product line development process that performs a model driven generation of agents embedded in heterogeneous smart objects with different degrees of self-management. However, once deployed, the home energy assistant system has to be able to evolve to self-adapt its decision making or devices to new requirements. So, in this paper we propose a model driven mechanism to automatically manage the evolution of multi-agent systems distributed among several devices.