A private LoRaWAN network is the one farm connectivity setup with no recurring bill. You buy one gateway, mount it high, run your own free network server, and every sensor you add after that talks over unlicensed spectrum with no SIM card and no monthly data charge per device. The cost is front loaded into the gateway, then it goes flat.
That is the whole reason to build your own network instead of buying cellular or NB-IoT sensors. If you already know what LoRaWAN is and why it fits scattered, low-data farm sensors, this piece is the next step: how to actually stand one up. If you are new to the technology, start with our earlier walkthrough on LoRaWAN sensor networks on the farm, then come back here to build.
There are two ways to connect a scatter of remote sensors across a place too big for wifi. The first is cellular or NB-IoT sensors. Each one is plug and play, but each one also carries its own SIM and its own recurring data charge. Ten sensors is ten little bills, forever. Add a moisture probe next spring and you have added another line item.
A private LoRaWAN network flips that math. You buy a gateway once, mount it high, and run a network server yourself for free. Every sensor after that is a one-time hardware cost with no per-device data plan. The recurring bill goes to zero because you own the whole stack instead of renting a slice of a carrier's.
Be honest about the tradeoff, because it is real. For one or two sensors, cellular is simpler and cheaper up front, and this whole build is overkill. The private network wins when you have many low-data sensors spread over ground a single gateway can reach - tanks, gates, soil probes, bin monitors, fuel tanks - and you are tired of watching per-sensor bills stack up. That is the buy signal. The cost is front loaded into the gateway and the work of mounting it, and it is flat from there.
The gateway is roughly 80 percent of your result, and antenna height is the single biggest lever on how much of your ground you cover. Everything else is refinement. Get the gateway high and clear, and a modest unit outperforms an expensive one mounted badly.
On hardware, you have a range. A starter indoor or light outdoor gateway runs in the low hundreds of dollars; a rugged, weather-sealed outdoor gateway built for a pole or tower costs more. Treat any dollar figure here as illustrative - prices move by vendor, region, and season, so get a real quote before you commit. What matters more than the price tier is the mounting.
Put the antenna as high and as unobstructed as you practically can. Best-practice deployments aim for something on the order of 15 meters above ground, positioned slightly above the structures around it so buildings and trees do not throw radio shadows. On a farm the natural spots are already there: the barn roof, a grain leg, a windmill or tower, a tall pole. Higher and clearer beats newer and pricier every time.
Antenna gain is where people quietly sabotage their own coverage. Do not just grab the highest-gain antenna on the shelf. A roughly 6 dBi antenna gives a balanced pattern - decent range plus coverage for sensors below and near the gateway. High-gain antennas flatten that pattern into a thin disc aimed at the horizon; they are for rural deployments where your sensors are far away and roughly level with the gateway. That is often the farm case, but not always. If half your sensors sit close and low, a tall high-gain antenna can shoot right over them. Match the antenna to your terrain, not to the biggest number.
The gateway needs one connection to the internet to pass your sensor data up to the network server. This is the honest asterisk on "no cellular bills": the gateway may still use a single cellular or satellite link if the barn has no wired internet. But that is one connection for the entire network, not a SIM per sensor. The savings are real; they are just per-sensor.
Wired ethernet is the best backhaul when you can run a cable or Power over Ethernet to where the gateway lives. It is stable, it is cheap once the run is in, and it does not fight weather. Wifi works if the gateway sits within reliable reach of the farm's existing network. Cellular or satellite is the fallback when the mounting spot has no other option - a gateway on a pole in a far field, for instance.
For any setup you genuinely cannot afford to lose, consider dual-path: ethernet as the primary link with cellular as automatic failover. That is more common on critical commercial installs than on a first farm build, but it is worth knowing the option exists before you decide the barn's single internet line is good enough.
Running your own network server is the step that removes the recurring cost, and you have two credible free paths.
The Things Stack Community, often called The Things Network or TTN, is the easiest on-ramp. It is free to start, well documented, and you can have a gateway registered and passing data in an afternoon. It is the fastest way to prove the whole idea works before you invest more.
ChirpStack is the cleanest fully-owned option. It is MIT licensed, fully open source, and self-hosted by default - typically as a set of Docker containers on a small always-on machine. It handles device activation, decrypts your sensor payloads, and manages gateways and tenants, with no vendor tier and no usage cap. You own everything and keep 100 percent of your data. For a farmer who wants the whole stack under their own roof, ChirpStack is the straightforward pick.
If "run your own server" sounds like a data-center project, it is not. Engineers have documented standing up a standalone private network on an off-the-shelf starter gateway - a Dragino LPS8v2 - with free dashboard software and no cloud account at all. An affordable gateway plus free software gives you a working, genuinely private network. That is the proof this is farm-scale work, not enterprise IT.
Add sensors using OTAA - Over-The-Air Activation - not ABP. OTAA is the more secure and more scalable join method, and it is the modern default. Get it set up once and adding your tenth sensor takes minutes.
Here is the mechanism in plain terms. With OTAA, the device ships without live session keys. You provision three identifiers into the sensor - the DevEUI, the JoinEUI (older docs call it the AppEUI), and the AppKey - and you enter the matching AppKey into your network server. When the sensor powers on, it negotiates fresh session keys with the server on each join, and the AppKey itself is never sent over the air. ABP, the alternative, burns fixed keys into the device with no join step. It is simpler to grasp but less secure and harder to manage once you have more than a couple of devices.
One practical tip that saves a frustrating afternoon: do the first join somewhere with guaranteed coverage, like your workshop or right next to the gateway, before you haul the sensor out to the far pasture. Confirm it joins and reports there, then move it to its real home. Debugging a silent sensor is a lot easier when you have already seen it work.
Run a coverage walk before you mount twenty sensors. It is the cheapest insurance you can buy against discovering the back forty sits in a radio shadow.
The farm-scale version of a professional RF site survey is simple. Power up the gateway in its final mounted position. Take one cheap node - a spare sensor or a coverage-test device - and walk or drive it to the corners of your ground that actually matter: the far tank, the bottom gate, the pasture behind the tree line. Watch whether packets keep arriving as you go. Where they stop, you have found a gap you can fix now, by raising the antenna, changing gain, or repositioning, instead of after everything is deployed.
Set your expectations at the conservative end. Optimal placement yields roughly a 2 to 3 km radius with solid signal penetration. Elevated gateways under near line-of-sight can reach much further - the brochures quote figures like 15 km or more - but treat that as a ceiling, not a plan. Design to a realistic single-digit-mile coverage figure and let the survey tell you the truth for your specific terrain, trees, and buildings.
Data arriving at your server does nothing until it reaches you. The path from "packets are arriving" to "a text when the stock tank drops below a level" is a solved, no-code to light-code job.
Your network server can push each sensor reading out two ways. It exposes an MQTT server you can subscribe to for uplinks, and it supports webhooks as the simplest way to shove each reading to an outside service. From there, Node-RED - free, visual, drag-and-connect - hooks into that MQTT feed, parses the payload, and drives a dashboard or fires an alert. Recent walkthroughs lay out the exact MQTT-to-dashboard flow step by step, so you are following a paved road, not inventing one.
A worked example makes it concrete. A float sensor on your water tank reports a level every few minutes. Node-RED reads that value, compares it to a threshold you set, and when the tank drops below it, sends a text to your phone. That is the payoff of the whole build - not a screen full of numbers, but the one message that saves you a trip out to a dry tank on a hot afternoon.
Build a private LoRaWAN network when you have many low-data sensors spread across ground a single well-placed gateway can cover, and you are done paying a recurring cellular bill on every one of them. That is the case this whole playbook is written for.
Skip it for one or two sensors - cellular is simpler and the gateway is not worth the effort. Skip it for anything that needs real bandwidth, like cameras or images; LoRaWAN moves tiny packets, and that is a hard physical limit, not a setting you can change. And go in clear-eyed that the gateway is genuine work: you have to mount it, power it, ground it against surges, and maintain it. The reward for that work is a network where the next sensor, and the one after that, costs nothing to run.
We keep a running library of practical farm technology and connectivity guides for exactly these decisions. If you are weighing your first build, the earlier LoRaWAN sensor networks explainer is the companion piece to this one - that is the "should I" and this is the "how to."
Buy one gateway, mount its antenna as high and clear as you can, and run a free network server yourself - either The Things Stack Community or self-hosted ChirpStack. Onboard each sensor with OTAA. The gateway needs one internet connection, but every sensor after that runs with no SIM and no monthly data charge, so the recurring per-device bill goes to zero.
Use The Things Stack Community if you want the fastest, easiest start - it is free and you can be running in an afternoon. Use ChirpStack if you want to own everything: it is MIT licensed, fully open source, self-hosted on your own hardware with no usage caps, and it keeps 100 percent of your data under your own roof. Both are free.
Mount the antenna as high and unobstructed as practical - around 15 meters is a common target, ideally above nearby buildings and trees. A barn roof, grain leg, or pole works well. Plan for a conservative single-digit-mile coverage radius rather than the 15 km headline figure, and confirm your real coverage with a walk-around survey before deploying every sensor.
Only for the gateway itself, and only if it has no wired or wifi internet. The gateway needs one connection to reach the network server, and that can be ethernet, wifi, cellular, or satellite. That is a single link for the whole network, not one per sensor. Wired ethernet is the best and cheapest option when you can run a cable to the gateway.
OTAA (Over-The-Air Activation) is the secure, scalable, modern default. The device negotiates fresh session keys on each join, and its AppKey is never sent over the air. ABP burns fixed keys into the device with no join step - simpler to understand but less secure and harder to manage once you have several sensors. Use OTAA, and do the first join near the gateway.
Join our list for practical guides on farm tech, precision agriculture, and tools that work.