Drones operating in a swarm can perform a range of different tasks and exchange information with one another. Up until now, civilian drone swarms are often controlled centrally via a mobile communications link. In the case of autonomous operations, this means that the individual drones are only able to communicate in a star topology via a central node – usually a mobile phone mast. If this node is damaged, however, for example as a result of a natural disaster, the swarm can no longer exchange information internally. This makes tasks such as searching for survivors more difficult, since the swarm can no longer operate cooperatively.

Mesh networks enable information exchange within the swarm

Researchers at the Fraunhofer Institute for Integrated Circuits IIS in Erlangen are working on a solution: “We’re developing a mobile ad hoc mesh network in which multiple drones are interconnected in a decentralized manner and communicate via Bluetooth, for example,”
explains Manuel Schrauth, Group Manager in the Communications Systems Division at Fraunhofer IIS. As a result, information exchange within the swarm no longer takes place in a star topology, but entirely decentralized – if necessary across multiple nodes. “This allows the
individual drones to continue communicating with one another even when no external communications link is available – a crucial advantage considering that only a fraction of the Earth’s surface has reliable mobile coverage.”

Highly dynamic decentralized networks have so far existed primarily in the military sector, where the technology is expensive, bulky and suitable only for heavy drones. No suitable mesh networks currently exist in the ultra-low-power sector. “Since the connection can run via Bluetooth, for example, our technology is extremely cost-effective and energy-efficient,” says Manuel Schrauth. “That means it’s suitable for the kind of small, lightweight drones that are typically used in civilian applications.”

The idea for developing ultra-energy-efficient ad hoc mesh networks emerged from projects within the GAIA initiative. Together with an interdisciplinary consortium, Schrauth and his team are working on satellite-based tracking devices to monitor vultures in Namibia that can communicate with one another in a decentralized manner. This allows the research team to analyze the animals’ behavior and draw conclusions about environmental changes, for example.

Continuous information exchange – even without mobile coverage

The team is now applying this knowledge to the networking of drone swarms by developing a table-based system. Now acting as nodes, the individual drones transmit at regular intervals what is known as heartbeats – referred to in technical terminology as advertisements. Schrauth compares this to a Bluetooth device signals for pairing and sends the message “I’m here” while searching for devices ready to pair.

As soon as two nodes detect one another, they update their respective tables, recording which other devices they have identified. If this process takes place every few seconds, it is possible to build up an almost complete overview of the entire network. Although each
individual node has only incomplete local information, a data packet can gradually propagate through the network, says Schrauth: “If node number one is looking for node number seven, it can pass the information on to the nearest point, node two, which then passes it on to node three, and so on. In this way, a data packet effectively hops from node to node until it reaches its destination, the required information being compiled iteratively.”

Task sharing through decentralized communication

As far as the user is concerned, communication within the swarm will ultimately feel like a point-to-point connection, according to Schrauth. Thanks to the new technology, multiple drones could in the future automatically launch from a docking case and form a predefined formation within the mesh network, thereby extending the swarm’s range and drawing on their different technical capabilities to create a three-dimensional map, for instance.

The researchers have already demonstrated the functionality of the technology in a simulation, and they now plan to transfer it to commercial hardware. Laboratory tests are currently under way using Bluetooth devices that process data across multiple nodes. The project group is expected to unveil a prototype of the technology next year. Manuel Schrauth and his team will present their initial findings at this year’s ILA.