24th International Conference on Embedded Wireless Systems and Networks

Featured Topic: Decentralized and privacy-aware intelligence in networks of resource-constrained heterogeneous devices.

Most current work on smart systems relies on AI and ML pipelines in which sensor nodes act primarily as data providers or as clients of centralised services. What happens when the intelligence moves into the network itself: when sensing, learning, inference and coordination are distributed across devices that differ widely in capability and that operate under tight budgets for energy, memory, bandwidth and heat? Decentralized and privacy-aware intelligence takes a different route: it emerges from decentralised interactions in the network itself, involving voting, consensus and local coordination, allowing data to stay with the contributor, as in federated systems and enabling levels of privacy that centralised architectures cannot offer.

This edition of EWSN looks into novel results that advance the platforms these systems run on, the protocols that hold them together, and the ecosystems of tooling, testbeds and measurement practice that make them reproducible. A consensus protocol behaves differently on a duty-cycled radio than on paper. A federated learning scheme that ignores thermal limits will not survive on a device worn on the face. The feature topic is therefore addressed to the systems layer, where the question is which function runs where, what the network has to guarantee for that split to hold, and what it costs in energy, latency and reliability.

The range of platforms making these demands keeps widening. Microcontroller-class sensor nodes and battery-free devices remain the core, joined by XR headsets and other wearables carried on the body, robotic platforms, software-defined vehicles whose functions are increasingly defined by software and updated over the air, micro-satellite constellations that must coordinate under severe power and intermittent-contact constraints, and, at the far edge of the spectrum, biocompatible nanodevices that sense and communicate inside living organisms. They differ in almost every respect, yet they pose the same design problem: the useful computation rarely fits on any single node, and the parts still have to stay in agreement.

The obstacles are real. Nodes are heterogeneous, resources are tight, and coordination is hard without a central authority. More capable components at the edge, at the telco or in the cloud can help orchestrate local interactions, provided privacy requirements are not compromised in the process.

EWSN 2027 invites workshop, tutorial and panel proposals across this space, from embedded and time-resilient ML to thermal-aware system design to molecular and bio-nano communication, and the keynote programme will reflect this focus.