Up front: choosing an IoT radio standard comes down to five axes - range, power draw, data rate, cost, and who owns the network. LoRaWAN is the right choice when many battery-powered sensors on your own site send small payloads and you want to control the network yourself - no SIM, no mobile contract. NB-IoT and LTE-M win when devices are mobile or widely scattered and connect through a carrier network. 5G (RedCap) and Wi-Fi are about bandwidth, not years-long battery sensors. This article compares the options axis by axis and shows the sovereign self-hosted path with LoRaWAN and ChirpStack.
Comparison table: LoRaWAN vs NB-IoT vs LTE-M vs 5G vs Wi-Fi
The key axes at a glance. Figures are typical real-world values and depend on hardware, environment and configuration.
| Criterion | LoRaWAN | NB-IoT | LTE-M (Cat-M1) | 5G RedCap | Wi-Fi |
|---|---|---|---|---|---|
| Spectrum | Unlicensed (ISM, EU868) | Licensed (carrier) | Licensed (carrier) | Licensed (carrier) | Unlicensed (2.4/5/6 GHz) |
| SIM / contract | None | SIM + subscription per device | SIM + subscription per device | SIM + subscription per device | None |
| Range | 2-15 km | up to ~10 km, deep indoor | ~5-11 km, mobile | cell-dependent (m-km) | ~30-100 m |
| Data rate | ~0.3-50 kbit/s | ~26 kbit/s DL / ~66 kbit/s UL | up to ~1 Mbit/s | up to ~150-220 Mbit/s | hundreds of Mbit/s to Gbit/s |
| Latency | high (Class A), low (Class C) | 1.6-10 s | 10-100 ms | tens of ms | a few ms |
| Battery life | years (up to ~10) | 10+ years | ~5-7 years | months-years | hours-days |
| Network operation | Your own (self-hosted) | Carrier | Carrier | Carrier | Your own |
| Mobility | Limited | Low | High (handover) | High | Local |
| Cost model | One-off gateway + ops | Per SIM/month | Per SIM/month | Per SIM/month | One-off AP + ops |
Range, indoor and outdoor
LoRaWAN and NB-IoT are both LPWAN technologies (Low Power Wide Area Network) built for long range (semtech.com). A LoRaWAN gateway covers 10 to 15 kilometers in rural areas, more like 2 to 5 kilometers in dense urban settings, while penetrating walls and basements well. NB-IoT achieves similar reach over the carrier network and excels at deep indoor penetration - smart meters in basements or sensors in manholes.
LTE-M offers comparable range but adds mobility with handover, the seamless cell switching needed for moving objects (vehicles, asset tracking). 5G RedCap and Wi-Fi are the shortest on range: RedCap depends on the mobile cell, and Wi-Fi reaches only tens of meters, aimed at local, mains-powered devices.
Power draw and battery life
This axis separates LPWAN from the rest. NB-IoT reaches 10 or more years of battery life in low-duty-cycle deployments, and LoRaWAN Class A sensors likewise run several years up to about a decade. Both sleep most of the time and transmit only briefly. LTE-M reaches roughly 5 to 7 years depending on usage.
5G RedCap and especially Wi-Fi draw far more power. RedCap is reduced versus full 5G, but it remains the worse choice for permanently battery-powered mass sensing. If you need years of maintenance-free measurement, you almost always land on LoRaWAN or NB-IoT. The mechanics behind it (spreading factor, duty cycle, device classes) are covered in detail in What is LoRaWAN?.
Data rate and latency
The order is clear: LoRaWAN is the narrowest at roughly 0.3 to 50 kbit/s - deliberately, as that is part of the range-and-power design. NB-IoT sits in a similar league at about 26 kbit/s downlink and up to 66 kbit/s uplink, but with high latency of 1.6 to 10 seconds. LTE-M offers up to about 1 Mbit/s and low latency of 10 to 100 milliseconds, so it also supports voice and reactive applications.
5G RedCap (also NR-Light, specified in 3GPP Release 17) closes the gap upward: up to roughly 150 to 220 Mbit/s downlink, with the stripped-down eRedCap variant (3GPP Release 18) capping at about 10 Mbit/s in both downlink and uplink. Wi-Fi plays in a different bandwidth class entirely, hundreds of Mbit/s to gigabit. Rule of thumb: the more data and the lower the latency you need, the further right in the table - and the higher the power draw and cost.
Cost: your own gateway vs SIM and mobile
The biggest structural difference is the cost model. Cellular IoT (NB-IoT, LTE-M, 5G) is billed per SIM per month. Examples as of 2026:
- Deutsche Telekom IoT: from around 0.99 to 2.95 EUR net per SIM per month depending on volume (Start to Business tier, b2b-tarife.de, April 2026).
- 1NCE Lifetime Flat: 12 EUR for 10 years with 500 MB and 250 SMS per SIM (1nce.com, October 2026) - cheap but volume-capped.
These amounts sound small, but they multiply with every device and run indefinitely. At thousands of sensors that becomes a predictable yet never-ending operating expense - plus carrier dependency.
LoRaWAN flips the model. There is no per-device fee, because no carrier sits in the middle. Costs are one-off for hardware - gateways from about 100 EUR (indoor) up to over 1000 EUR (rugged outdoor units), sensors around 20 to 100 EUR each - plus running the network server. That server, ChirpStack (MIT license, currently v4.19.2 from September 2026), is free (chirpstack.io). So the real cost driver is not licenses or SIMs but servers and operations - and that stays flat regardless of device count. More on operating models in IoT self-hosted vs. cloud.
Who owns the network? Sovereignty and scaling
This axis is often overlooked but strategically decisive. With NB-IoT, LTE-M and 5G the network belongs to the carrier. You rent connectivity, accept its coverage, tariffs and roadmap, and pass data over third-party infrastructure. With LoRaWAN and Wi-Fi the network is yours: you decide locations, capacity and data flow.
For scaling this means: cellular IoT scales linearly in cost - every device adds another SIM fee. LoRaWAN scales by area through additional gateways, each carrying very many devices while per-device cost trends toward zero. That is exactly the sovereign lever: no cloud lock-in, no per-device license explosion, full data sovereignty in the EU.
Decision guide: when to use what
- LoRaWAN: Many battery-powered sensors on your own defined site (campus, plant, building, district, field), small payloads, long life, your own network. First choice for sovereign, cost-effective mass sensing.
- NB-IoT: Widely scattered, stationary devices in deep indoor positions (smart metering, manholes) where your own gateways are impractical and carrier coverage exists.
- LTE-M: Mobile assets that need handover and low latency (vehicle, logistics and wearable tracking), optionally with voice.
- 5G RedCap: Medium to high bandwidth without full 5G overhead (industrial cameras, routers, demanding wearables); carrier rollout starts in 2026 in cities.
- Wi-Fi: High bandwidth at short distance for mains-powered devices (local gateways, video, fast data transfer).
In practice platforms combine several radio paths: LoRaWAN for mass sensing, MQTT/Wi-Fi for local gateways, occasionally cellular for mobile nodes. How these layers interact is shown in IoT architecture layers.
Sovereign: self-hosted LoRaWAN with ChirpStack
The decisive advantage of LoRaWAN is network ownership. Instead of renting connectivity per SIM, you run your own LoRaWAN network: your own gateways and a self-hosted network server with ChirpStack. ChirpStack is open source (MIT), merges network and application server into one application since v4, and ships MQTT, database and cloud integrations. It runs on an ordinary server with PostgreSQL, Redis and an MQTT broker.
This keeps full data sovereignty in the EU, charges no SIM and no per-device fee, and frees you from carrier tariffs. This open source model - from sensor through gateway to dashboard - is exactly what we build and operate at WZ-IT.
Build your own sovereign LoRaWAN network? We plan, build and operate self-hosted LoRaWAN networks with ChirpStack on your infrastructure - no cloud lock-in, no per-device license, GDPR-compliant from Germany. See our LoRaWAN service - View ChirpStack expertise - Book a meeting
To go deeper: What is ChirpStack? explains the network server in detail, and our IoT services show how we deliver complete platforms from sensor to dashboard.
