Georgios Exarchakos is an assistant professor on smart networks at Eindhoven University of Technology. Together with a team of young researchers, master and bachelor students he has created Gravelnet.

14 September 2018

Even advanced simulators could not reliably validate IoT innovations generated at the TUE department of Electrical Engineering. So they decided to create an open facility of their own.

IoT engineers need to make choices on the sensing, actuation and connectivity hardware, the communication protocols, platforms and middleware before implementing their solution. Yet, no facility out there gives them the flexibility to simultaneously integrate, prototype and validate all these choices. Sparked out of own projects, researchers at the Eindhoven University of Technology (TUE) decided to build one themselves, with the intention to significantly shorten the development cycle of integrated IoT solutions and improve their reliability.

The result is Gravelnet, a distributed testing facility for hardware and software systems in the IoT era. It helps a designer of distributed embedded systems and applications validate their behaviour in dense setups. Amongst other things, the facility allows for easy deployment of hardware components on an evaluation board, runtime configuration, and quick testing of communication modules, protocols and network QoS parameters in a real-world environment. For educational purposes logging and various application complexities can be hidden.

TUE Georgios Exarchakos 02
Georgios Exarchakos worked with a team of young researchers, master’s and bachelor’s students to create Gravelnet. Photo: Bart van Overbeeke

Monitoring nodes

At its core, Gravelnet is a network of monitoring nodes, so-called gravels. Any approved device can be plugged into a gravel via USB and may output binary or textual, structured or unstructured data to that interface. Gravelnet timestamps all incoming data and collects it in a dataset per experiment, accessible through a single portal.

Around this core, the TUE team has built a number of features allowing a variety of use cases. Registered users, once granted the necessary permissions, can remotely configure and manage participating devices. They can create, launch, schedule and stop experiments on a selection of the available gravels and their connected devices. They can view and download data from past experiments, create new and save configurations as default. They can build at runtime an IPv6 pipe between any terminal sitting outside the network directly to a device plugged to a node of their choice. They can build and deploy approved gravels and devices at any location to allow full control of experiments at their own sites.


Device lifecycle management for fleets of IoT devices

Microchip gives insight on device management, what exactly is it, how to implement it and how to roll over the device management during the roll out phase when the products are in the field. Read more. .

Data can be collected in the field without connectivity; it will be automatically pulled in once the connection is re-established. All gravels communicate over VPN with the portal and their software is automatically updated to ensure best privacy and data integrity practices. The code is open sourced to ease customizations, support of more hardware and firmware as well as new sites.

100 gravels

Gravelnet is a federated facility. The first site is located at the TUE’s Electrical Engineering building, which is also the home of the portal and management system. So far, the facility consists of a hundred fixed gravels, built upon Beaglebone Black Rev C boards, supporting eighty Telosb and twenty JN5168 edge devices. It’s scheduled to open late 2018, together with the CWTE Research Retreat.

Since its early stages, Gravelnet has been supporting research on smart resource allocation for reliable wireless edge networks. With the increasing density of wireless edge devices, resource (eg frequencies, timeslots, transmit power) overprovisioning is unable to support reliable and energy-efficient communications. The facility has helped at prototyping massively distributed smart adaptive resource-scheduling algorithms.

Redacteur: Nieke Roos