Blog
LPWA technologies

LTE-M vs NB-IoT: a technical guide to choosing

They are sibling technologies, but not interchangeable. One is for moving things that talk occasionally; the other is for stationary sensors that rarely say anything. Picking wrong multiplies power draw and retransmissions.

April 10, 20269 min

LTE-M (also called Cat-M1) and NB-IoT (Cat-NB1/NB2) are two 3GPP cellular profiles designed for low-power, low-cost devices. Both share LTE spectrum and both are part of 5G as the mMTC (massive Machine-Type Communications) component. But their capabilities differ enough that confusing them gets expensive fast.

Differences that matter in production

  • Bandwidth. LTE-M uses 1.4 MHz (up to ~1 Mbps DL on Cat-M1, ~3 Mbps on Cat-M2). NB-IoT uses 180 kHz (~60–120 kbps). If you send more than a few KB per day, LTE-M breathes; NB-IoT doesn't.
  • Latency. LTE-M: 100–150 ms typical, acceptable for live telemetry. NB-IoT: 1.5–10 s in connected mode, with spikes when the device wakes from PSM. NB-IoT is not fit for voice or for actuators that must react to a human command.
  • Mobility. LTE-M supports handover between cells: if the device rides on a truck, shared bike, or personal tracker, it's the only reasonable choice. NB-IoT originally only did cell reselection in idle; Release 14 added optional handover, but in practice most deployed networks still don't support it. For moving nodes, assume NB-IoT isn't viable.
  • Voice. LTE-M supports VoLTE. NB-IoT has no voice. If the use case involves an alarm call (elevators, eCall, senior care), NB-IoT is ruled out.
  • Indoor penetration. NB-IoT wins: thanks to repetitions and redundant coding it reaches MCL ≈ 164 dB (around 20 dB better than standard LTE), which translates to water meters in basements or buried sensors. LTE-M gets to ~156 dB. For difficult fixed assets, NB-IoT is the answer.

Power consumption

Both support PSM (Power Saving Mode) and eDRX. In a typical profile of one 200-byte transmission per day, a well-designed device can run around 7–10 years on two AA lithium cells with NB-IoT, versus 4–7 years with LTE-M. The difference comes from modem airtime: NB-IoT spends more time transmitting but at lower instantaneous power; LTE-M transmits faster but drives the PA harder.

If the project depends on 10-year autonomy, measure real consumption with the modem you will ship, on the carrier you will use, at the physical pilot site. Datasheet figures are indicative; field data rules.

Geographic availability

In Europe, most tier-1 carriers have both networks, with nuances: Vodafone has historically pushed NB-IoT across its European footprint, Orange blends both, Deutsche Telekom has excellent NB-IoT coverage in Germany and the Netherlands. In the U.S., AT&T and Verizon have partially shut down their NB-IoT networks and operate primarily LTE-M. In Asia, Japan, South Korea, and China went heavy on NB-IoT.

For a device that will ship globally, today's only reasonable answer is: modem with LTE-M + NB-IoT + 2G/3G fallback where still available, on a SIM that roams on both profiles. Force NB-IoT only and you give up the U.S. Force LTE-M only and you lose strong coverage in Germany, the Netherlands, and much of the Asian market.

Roaming and regulation

Here is a silent trap. Many European networks let an LTE-M module roam as plain LTE — the visited operator assigns it a legacy LTE bearer and done. But NB-IoT roaming relies on explicit carrier-to-carrier agreements, and those agreements aren't universal. Today, if your SIM comes from a Spanish MNO and your device travels to Germany, NB-IoT roaming isn't guaranteed. Ask your provider for the actual NB-IoT roaming matrix, country by country.

How to choose: practical decision tree

  • Does the device move? → LTE-M. NB-IoT with handover only if the carrier confirms it in writing.
  • Does it need voice or an alarm call? → LTE-M (VoLTE).
  • Does it send commands expecting a reply under 2 s? → LTE-M.
  • Is it a fixed sensor in a basement, buried, or behind concrete, sending a few bytes a day? → NB-IoT.
  • Is it a water/gas/electricity meter with 10-year autonomy as a hard requirement? → NB-IoT, unless deploying only in the U.S.
  • Shipping globally with no SKU fragmentation? → dual LTE-M + NB-IoT module and a multi-carrier SIM that supports both.

Common mistakes

  • Testing LTE-M in the office and assuming NB-IoT will behave the same. It won't. Repeat the tests with NB-IoT at the real location — penetration profiles differ.
  • Forgetting that 2G/3G sunset breaks your fallback. If your module depends on 2G as backup, and the device route crosses countries that already shut down 2G (Switzerland, Australia, Japan), your fallback doesn't exist.
  • Assuming NB-IoT roaming. Ask for the list of countries with an active agreement and the date of last validation.

At iot.cards we sell multi-carrier SIMs that support LTE-M and NB-IoT wherever an agreement is in place, with automatic fallback. If you're preparing a large-scale rollout, it pays to test two or three carriers in the real service area before signing.

Working on a project like this?

Drop us a line and a network engineer gets back to you in under 24 h.

Talk to an engineer