"First mile" data extraction with LoraWAN     Fri Aug 6 2021
For real-world AI applications, your ability to access more relevant data is just as important as your ability to deploy sophisticated machine-learning models. That's why at Sense6, we also specialize in some IOT-related technologies (MQTT, Wireguard, deploying cool sensors). So what does this have to do with LoraWAN?

The situation: a lot of use-cases need to react in real-time to sensor data. To do that, you need to quickly get the data out of the sensors and onto a secured network in order to analyze them. If your sensors are indoors and have access to WiFi, or they're on some fancy robot and you don't mind adding a GSM cellular connection (and have the money and electric power to spare), then maybe LoraWAN is not for you. But in other cases things are not so simple: It's hard to get the data out of the sensors, what we mean by the "first mile".

The problem: imagine a logistics company wanting to monitor shipping containers, or a farmer wanting to survey crops in open fields, or a company wanting to monitor its office room temperatures but without the resources to add internet everywhere.
In these cases, it's not practical to use traditional communication protocols, usually for 3 reasons:
  • Too costly: wiring up a warehouse or an office with Ethernet requires significant investment
  • Low reach of WiFi: while WiFi would be great, its range is a few dozen metres indoors and maybe a couple hundred metres outdoors. Our use-cases require hundreds of metres indoors and kilometres outdoors
  • Power constraints: most of these nodes need to run outdoors and therefore on battery. We can't be recharging batteries every week, nodes must be power efficient

A solution: LoraWAN. In a nutshell, LoraWAN enables you to connect cheap sender- and receiver-devices across many kilometres outdoors and hundreds of metres indoors. These Lora devices are cheap and can run for many months on a single battery charge (we didn't test this yet). There is a trade-off however, you are very restricted in the amount of data you can send. Forget sending images or even long texts, best is to restrict Lora messages to a few numbers. Tough, but not a showstopper: a heat sensor sensing a fire doesn't need to send a newspaper article, 1 bit could suffice (fire = 1 or something like that).

It's not glamorous, but we've got a strong signal here... and we think that's amazing.

Our first experiment: we tested prototype #1 during a storm in the countryside: strong rain and howling winds (not very apparent in the photo) couldn't keep us indoors. Right away the "sender" node (half the size of a credit card) built a strong connection with a "receiver gateway" (size of a pocket book) more than 1 kilometre away despite the receiver being indoors and a few trees and houses separating us. So cool!

We'll be testing the stability of these devices in the months to come. I think you can tell we're excited about this technology, it opens more doors for powerful AI applications.
Click here to show all blog posts
Powered by Sense6