Localization is an important subject of the Internet of Things (IoT). However, GPS localization uses a lot of battery and cannot be used in most low-power applications. For this reason a solution based on low-power transmission such as LoRa can be developed. A first estimation can be obtained using the Swisscom LPN LoRa network, but more precision is sometimes needed, for example in the case of search and rescue of avalanche victims.

We can imagine a small wearable chip that can be triggered in emergency mode when required that would send "locate-me" signals. In order to increase the precision of the estimation, mobile gateways can be deployed in the vicinity of the beacon estimated position. During my master project done at the Swisscom Digital Lab, I developed a solution for localization using drone-mounted mobile gateways. The main idea is to use the high mobility of the drones to locate the beacon. Based on the signal strength of the transmission, a distance estimation can be computed between the drone and the beacon. By moving into different measures in different positions, the drone can compute the beacon's position.

Localization Drones.gif

The animation above shows the application with three drones. The first drone waypoints are based on the position estimate given by the Swisscom LPN network. Multiple estimations in different positions can be made to increase the precision. Using three drones a precision below 10 meters was obtained in less than 5 minutes in simulations.

We performed several tests with one or three drone in Sion at GamSion testing ground. There, the first flight tests of the solution with one drone computed a precision around 30 meters. The entire search mission took around 6 minutes.

The second flight test with further improvement of our algorithm lead to a 15m precision in localizing the node!

This result is already satisfactory for implementing in real life, event thou improvements are always possible.

Helping Drones to See
he collaboration with Swisscom gave rise to the pilot trial, which got underway in January 2018. The seven mounted sensors provided the desired information about the airspace, meaning that Involi now has something concrete to show to interested parties: a 3D airspace model with accurate and reliable data. Meanwhile, the sat nav is still not showing the antenna site near Premier as a point of interest.