Middleware for CPS, IoT and Edge Computing

In this topic, the team is considering (i) publish-subscribe middleware technologies for interoperability of IoT/Edge devices, (ii) the specification and provisioning of quality-of-service attributes to the supported data streams, and (iii) efficient data interoperability in smart cyber-physical systems.

Research activities address both the specification and provisioning of Quality-of-Service attributes to the supported data streams, as well as the support to efficient data interoperability in smart cyber-physical systems.

The notion of Quality-of-Service contracts and Service Level Agreements is of paramount importance to provide applications with the required data processing resources; however, this support should be as much as possible transparent to users in what concerns the mapping to the underlying system resources and parameters. At the same time, middleware technologies need to cope with performance requirements on top of heterogeneity and resource limitations of IoT and edge devices.

Management of computation in parallel and distributed IoT/Edge ecosystems

In this topic, the team is tackling approaches and technologies for (i) mapping and scheduling of concurrent and parallel computation in edge devices, (ii) mapping and scheduling of workflows in the edge-cloud compute continuum, (iii) and monitoring of computational resources.

Work focuses on approaches and runtimes for the mapping and scheduling of concurrent and parallel computation, both at the level of workflows in the edge-cloud compute continuum, as well as local computation in heterogenous edge platforms.

With respect to the increased flexibility and openness requirements of smart applications, it is expected that the usage scenarios will dynamically change during execution, requiring that the scheduling approaches adapt at run-time, considering both reactive and intelligent approaches.

And to support these dynamic adaptation capabilities, an important work area is resource monitoring, where activities have been focusing in online monitoring of both time and energy properties.

Development of cyber-physical systems and software

In this topic the team considers (i) AI at/for the edge, (ii) Model-based development of embedded software, (iii) security of IoT and edge devices and (iv) Concurrent and parallel software.

One main area of activity is the use of AI techniques to manage the computation and orchestration at the edge (and in the edge-cloud compute continuum), as well as the use of AI algorithms in resource constrained devices at the Edge. A particular area of application is AI-based cybersecurity at the Edge.

Another main areas of activity is the reliable development of concurrent and parallel software. In parallel, the team is also addressing approaches and tools for the early analysis of software execution from models, allowing to reason on fulfilling non-functional properties of applications, such as time and energy.