The Things Network, the crowdsourced IoT lab

Yesterday I went to the second meet-up of The Things Network Paris. For those who don’t know it , The Things Network (TTN) is a community-driven and crowdsourced IoT network. Anybody can extend the network coverage, by setting up its own gateway, as long as it is compatible. This network relies on the popular LoRaWAN technology. It ambitions to boost innovation by providing free IoT connections, to anyone.

This initiative originated from Amsterdam, but has since expanded to over 40 countries. Around 200 communities now build local coverage for the crowdsourced network. The biggest local network is currently in Zurich, with over 30 gateways. You can check the full story of TTN on their website. I will detail my thoughts about the whole initiative in another post, soon.

Yesterday's meet-up was organized by the still small, but motivated Paris community. With 5 operational gateways (7 in the agglomeration), the network already enables some use cases. But it is still far from competing with professional alternatives in terms of coverage. Especially when you consider operators like Sigfox, which target international coverage.

While still in its infancy, TTN allows for some interesting observations on IoT networks. Renaud Lifchitz exposed some of them during the meeting. He's the security researcher behind the recent revelations on LoRa and Sigfox security flaws. The observations are based on a sample of the global packets exchanged on TTN. Yet it is hard to say if the sample is statistically representative. Here are the figures he presented:

  • 70% of the 'join' packets are sent on the 1st channel (for example 868.1 MHz in the 868 MHz band). They are supposed to be equally distributed on the available channels, for better efficiency.
  • 50% of the packets are sent with the configuration allowing for smaller ranges (SF7 for the tech savvy). 25% are sent with the configuration allowing for the longer ranges (SF12). Anything in between doesn't seem to be much used. These configurations should however be dynamically changed to optimize transmissions (the LoRa tech includes an adaptive data rate).
  • 2% of packets require an acknowledgement. So much for the heated debate about bidirectional communications...

Hopefully The Things Network will provide official statistics on a regular basis in the future. This would help a lot to understand how people use such IoT networks. It would also help spotting inefficiencies or improper implementations (probably responsible for the first two points).

All in all, TTN already allows for some interesting analysis. It is also useful for people to connect their prototypes or personal devices for free, thus spurring local innovation. While it doesn't seem to be ready to compete with professional networks, this kind of network nevertheless raises serious questions about the market and the regulation, that I will address in another post. Stay tuned!