LoRaWAN with ProRF and The Things Network

Pages
Contributors: Nick Poole
Favorited Favorite 10

What is LoRaWAN?

TL;DR: LoRaWAN™ is like cellphone towers for IoT that allow battery powered things to talk to the internet.

If you're interested in the Internet of Things, you've probably heard of LoRa™ radios. LoRa™ is a fancy new kind of FSK (Frequency-shift keying) modulation that was developed by Semtech Corporation, it's short for Long Range and it lives up to the name.

Because LoRa™ is just a modulation scheme, it can be used anywhere that you would use other types of packet radio. In fact, one of our favorite LoRa™ modules — the HopeRF RFM95W — is a direct drop-in replacement for their standard FSK radio module — the RFM69HCW — which makes it easy to upgrade your existing packet radio projects to take advantage of Semtech's long-range, low-power technology.

One particular use of LoRa™ is uniquely exciting, though: LoRaWAN™

LoRaWAN™ is a network standard developed and maintained by the LoRa Alliance: an open association of collaborating members — mostly large tech companies — that's designed to allow low-power devices to connect to each other and the internet using public gateways. The standard dictates that devices can move freely between gateways. Essentially, it's like cellphone towers for IoT. The difference is that anyone can inexpensively own and operate a LoRaWAN™ gateway, so building the network is easy. And because anyone could theoretically operate a network server, it's relatively resistant to monopoly.

The LoRa Alliance has laid out a template for LoRaWAN™ networks and this template consists of 4 parts:

  • Nodes
  • Gateways
  • Network Servers
  • Application Servers

Nodes

A LoRaWAN™ node is an endpoint device, such as a sensor or an actuator of some kind. A node can be up to 10km (6 miles!) away from a gateway in ideal conditions with the right radio modules. The connection between a node and a gateway is very low bandwidth — between 0.3 and 50 kbps — but it is bi-directional. Nodes need to be smart enough to encrypt and decrypt packets, handle network authentication and respect the duty cycle (we'll talk about that in a minute) but these are tasks that can be easily achieved on super low-cost microcontroller hardware. The only other thing that a node needs to have is a LoRa™ radio and some kind of antenna.

Gateways

A LoRaWAN™ gateway — sometimes called a concentrator — is kind of like a cross between a cell tower and a WiFi router. It's the bridge between the nodes and the internet. Because of the long range capabilities of LoRa™, a single gateway can theoretically service entire cities or hundreds of square kilometers. Ideally, however, a given node will be "heard" by multiple gateways to ensure the best network fidelity. Nodes don't intrinsically know when they've been "heard" they just scream into the void, so it's always good to have multiple gateways within range. If multiple gateways happen to get copies of the same packet, that will be taken care of upstream at the network server. Besides range, another limitation to any LoRaWAN™ gateway is the number of channels it has. The number of channels that a gateway has is the number of nodes that it can talk to at once. At first, this seems extremely limiting since many consumer gateways are around 8 channels, but then we introduce the concept of duty-cycle.

Nodes agree to adhere to a duty cycle limitation. This limitation is actually enforced by government regulation in many parts of the world in an effort to keep the airwaves open for everyone to use. A duty cycle is a measure of the fraction of time that a resource is in use. For instance, if you're transmitting on a particular frequency for 2 seconds straight every 10 seconds, you're operating at a 20% duty cycle. So what's the duty cycle limitation for a LoRaWAN™ node? It depends on your local laws, but it's probably 1%. For certain applications it may be as low as 0.1%, but this isn't as low as it sounds. A 1% duty cycle represents almost 15 minutes of combined airtime per day, which far exceeds the fair usage policy of most free networks. And when you consider that a packet takes tens of milliseconds of airtime, these restrictions feel a lot more permissive. Also, by adhering to this duty cycle, we increase the number of nodes that can be serviced by a single gateway by a hundred times!

Network Servers

LoRaWAN™ nodes don't know anything about the internet so the gateway can't just turn LoRaWAN™ packets loose on the web and hope for the best. There needs to be a particular server that expects those packets and knows how to deal with them. This is the network server and — as the control center of the network — it has a lot of jobs to do. Its primary job is to direct packets between gateways and application servers. Since LoRaWAN™ allows for uplinks (messages to a server from a node) and downlinks (messages to a node from a server). Because the network server controls all gateways on the network, there are a lot of packets to juggle.

Another thing that the network server does is to de-duplicate packets coming in from multiple gateways. Since any node might be within range of multiple gateways and there's no hand-off when moving between them, the packets just get duplicated as each gateway sends its copy to the network server. The network server compares them and throws out identical packets. Finally, depending on the capabilities advertised by the network operator, a network server might be doing all kinds of other things like monitoring airtime usage and managing subscriptions. Some networks even offer localization for all nodes, triangulating them using "Differential Time of Arrival" techniques with multiple gateways.

The big thing to remember about the network server is that it does all of the behind-the-scenes work that makes a LoRaWAN™ network operate. Just like in the early days of cellular networks, not all gateways talk to the same network. LoRaWAN™ is only a network standard, not a network in and of itself. On top of this, the network server probably isn't going to run any application-specific code to talk to your devices, but it will know where to send your packets so that your application-specific code can see them...

Application Servers

An application server is a server that's connected to the network server (usually somewhere on the internet) that knows specifically what to do with packets from a given node or type of node. For instance, if you have a website that shows the current weather conditions at a certain LoRaWAN™ weather station, the weather station is sending packets which are being relayed to the network server by one or more gateways. The network server isn't interpreting that weather data or serving you the website... that stuff is happening on an application server. The application server and the node are both registered with the network server, so it knows to send packets from the weather station node to the weather station application server. Application servers can be anything from an IFTTT Webhook to a Raspberry Pi somewhere on the web. Sometimes an application server makes data available to browsers, sometimes it simply manipulates data and sends it back out to nodes over the network server.

"Okay, So How Do I Get In On This?"

There are a lot of LoRaWAN™ networks all over the world, some free to use and some paid. The largest free network today seems to be The Things Network, and it's growing every day. Even if you're not currently covered, you can set up your own gateway and join the network to help it grow!