Core Technology
LoRaWAN
The radio network underneath most of what we deploy. It trades bandwidth it doesn't need for range and battery life it does — which is why one gateway can cover a whole building or a whole district, with no cable to pull and no monthly plan to pay.
01 /The Idea
It's a Trade, and It's the Right One
LoRaWAN isn't a faster radio. It's a deliberate decision to give up things a sensor never needed, in exchange for the two things that actually decide whether a deployment happens.
What it gives up
- Bandwidth. A LoRaWAN message is a few dozen bytes — a temperature, a pulse count, a tank level. Never a photo, never video, never a file.
- Speed. Devices talk in short bursts and sleep in between, so a reading arrives seconds after it's taken, not milliseconds.
- Chattiness. Devices report on a schedule or on an event, not continuously. That's a design constraint, not a defect.
What it buys
- Range. The same message that dies in a hallway on Wi-Fi crosses a campus, several concrete floors, or a whole district.
- Battery. Radios that sleep almost all the time run 5–10 years on a single cell — and some devices outlast that.
- Cost per point. One gateway serves thousands of devices, so the hundredth sensor costs roughly what the first one did.
A water level, a pulse count, a door state, a temperature — none of them are bigger than a text message. Sizing a network for video when all you're moving is numbers is what makes monitoring expensive.
02 /Mechanics
How a Reading Actually Gets to You
Four steps from a sleeping device in a basement to a live point on your map.
03 /Comparison
Measured Against Everything Else
Wired fieldbus, wired IP, Wi-Fi, and cellular all have deployments where they win outright. This is where each one actually lands when the job is many small sensors over a wide area.
Scroll to compare →
| LoRaWANUnlicensed sub-GHz | Wired FieldbusBACnet MS/TP, Modbus RTU | Wired IPBACnet/IP, Ethernet, PoE | Wi-FiThe building's WLAN | CellularLTE-M, NB-IoT, CAT-M | |
|---|---|---|---|---|---|
| Adding the next sensor | Bracket or magnet, register its keys, done — minutes, no cable. | Extend the daisy chain: conduit, terminations, an electrician per point. | A home run back to a switch, plus a spare port and a ceiling drop. | Quick to join — but it needs mains power and an IT-approved SSID. | Easy to mount, then activate a SIM and start paying every month. |
| Airtime cost per device | None. Unlicensed spectrum on a gateway you own. | None. | None. | None. | A data plan per device, per month, for the life of the device. |
| Power source | Battery, as the normal case. 5–10 years is unremarkable. | Bus-powered or local mains. | PoE or mains at every single point. | Mains, effectively always — Wi-Fi radios drain cells fast. | Battery is possible, but each message costs far more energy. |
| Coverage per piece of infrastructure | 1–3 mi across a city, 9 mi+ in open country — from one gateway. | About 4,000 ft of bus total, and every foot of it must be pulled. | About 330 ft per Ethernet run, then you need another switch. | A floor or a few rooms per access point. | Wherever the carrier already has coverage — and nowhere else. |
| Devices per gateway | Thousands on one 8-channel gateway. | Tens per segment before you need a repeater. | One switch port per device. | Dozens per AP before throughput suffers. | Not applicable — every device is its own subscriber. |
| Does it touch your IT network? | No. A separate radio network; only the gateway has an uplink. | No, but it's bound to the building automation controller network. | Yes — every sensor becomes a managed device on your LAN. | Yes — every sensor is on your WLAN, with credentials to rotate. | No, but each device sits directly on a public carrier network. |
| Bandwidth | Kilobits. Readings and counts — never files or video. | Enough for polling and control loops. | Effectively unlimited. This is where video belongs. | High, wherever the signal is strong. | Megabits on LTE-M — far more than telemetry needs. |
| What it's genuinely best at | Many battery sensors, spread wide, reporting small numbers often. | Controlling equipment that is already wired. | Cameras, controllers, and anything moving real volume. | People's devices and mains-powered gear near an AP. | Assets that roam beyond any gateway you own. |
None of this is an argument to rip anything out. BACnet and Modbus still run the equipment they were built for, and we read them where your building already speaks them. The point is narrower: a new measurement point no longer requires new wire, a new switch port, or a new line on a carrier bill.
04 /Economics
Three Costs That Simply Stop Existing
Not reduced — removed. Each one is a reason monitoring projects get scoped down to a fraction of what was worth measuring.
The sensor is rarely the expensive part. Reaching it is.
Cellular telemetry makes every sensor a subscriber account.
Battery life is an operations budget in disguise.
05 /Indoors
Why It Wins Inside a Building
The hard part of a smart building was never the sensors. It was that measuring one more thing meant a contractor, a ceiling, and a change order.
06 /Outdoors
Why It Wins in the Field
Public works assets share a profile that defeats almost every other radio: scattered across miles, often below grade, rarely near power, and expensive to visit.
07 /Security
Its Own Network, Encrypted End to End
Security here comes mostly from architecture rather than configuration — there are fewer ways in because there is less exposed to begin with.
For facilities and agencies with a security review to clear, this is usually the shortest conversation of the project: the sensors are not on the network IT is responsible for, they hold no data at rest worth taking, and they cannot be reached from the outside.
08 /Limits
Where LoRaWAN Is the Wrong Tool
It's a narrow radio, and pretending otherwise is how deployments fail. Here's what we deliberately put on something else.
Every deployment we run mixes at least two of these. LoRaWAN carries the many small things; cameras and wired systems carry the few heavy ones. They all land in the same platform, on the same map, on the same clock.
09 /Our View
Why We Think This Is the Future
For thirty years, deciding what to measure was really a decision about what you could afford to wire. Every candidate sensor had to survive a cost review that had nothing to do with its value and everything to do with its distance from a cable. Most didn't survive it. That's why buildings are full of systems nobody has data on, and why cities discover a failure when a resident calls.
What LoRaWAN changes is not the sensor — it's the marginal cost of the next one. When coverage is a one-time investment and a device runs for a decade on a battery, measuring one more thing costs roughly what the device costs. Below that threshold, you stop rationing. You instrument the whole storm system instead of the three outfalls that made the budget, and the whole building instead of the floor with the complaint.
That's the shift we're building on: measurement stops being a capital project and becomes something you can simply add. Everything else we make — the live map, the alerts, the data lakehouse that keeps years of it — assumes a world where you finally have data from everywhere, rather than from the handful of places that were cheap to reach.
Bring Your Own Gateway, or Let Us Build the Coverage
Already running LoRaWAN? Point it at Gostly and your existing sensors show up on the map. Starting from nothing? We'll survey the coverage and stand it up.
