NB-IoT vs LTE-M: how they differ, how to choose and what to test

LTE-M and NB-IoT are both 3GPP low-power cellular technologies for IoT, introduced in Release 13. LTE-M uses a 1.4 MHz channel and offers higher data rates, millisecond-range latency, full mobility with handover, and optional VoLTE. NB-IoT uses a 200 kHz channel and trades speed and latency for deeper coverage and simpler modules, which suits stationary sensors. In practice, the right choice also depends on what each operator and roaming partner supports, so both technologies have to be tested on the live networks where devices will run.

NB-IoT vs LTE-M at a glance

	LTE-M (Cat-M1)	NB-IoT (Cat-NB1 / Cat-NB2)
Channel bandwidth	1.4 MHz	200 kHz (one 180 kHz resource block)
Typical peak data rates	Around 300 kbps down / 375 kbps up for common half-duplex modules	Around 30 kbps down / 60 kbps up (NB1); up to about 127 kbps down (NB2, Release 14)
Latency	Milliseconds range in normal coverage	Seconds possible; 3GPP’s design target for exception reports was within 10 s
Mobility	Full mobility, including connected-mode handover	Designed for stationary devices; limited mobility support
Voice	VoLTE possible where the operator deploys it	No voice
Coverage	3GPP target of 155.7 dB maximum coupling loss; up to 164 dB under GSMA analysis with CE Mode B	164 dB maximum coupling loss target
Coverage enhancement	CE Mode A (mandatory, up to 32 repetitions); CE Mode B (optional, up to 2048)	Repetition-based coverage enhancement
Power Saving Mode	Yes; sleep (T3412 extended) up to about 413 days	Same
eDRX cycle	Up to 43.69 minutes	Up to 10,485.76 seconds (about 175 minutes)
Roaming	Agreed per partner; generally more widely available	Agreed per partner; narrower footprint in many markets

Sources: data rates from Nordic Semiconductor. The latency target is from 3GPP TR 45.820. Coverage figures come from the GSMA’s LTE-M coverage analysis. CE modes, PSM and eDRX values are from the GSMA LTE-M and NB-IoT deployment guides.

A few points behind the table:

• Throughput is mostly about what the application can do. NB-IoT is fine for a few hundred bytes a day. A firmware image of several megabytes, or a camera snapshot, is a very different load, and LTE-M handles it much faster.
• Coverage figures are design targets. Real coverage depends on deployment, band, device power class and whether the network enables the deeper coverage modes. The GSMA LTE-M guide notes that a roaming device reverts to CE Mode A if the visited network has not enabled CE Mode B.
• Voice is optional on LTE-M. The GSMA LTE-M guide does not make VoLTE a mandatory part of the baseline, so a voice-capable device needs VoLTE confirmed on each network.

How to choose between them

Requirement	Points toward
Device moves (vehicles, trackers, wearables)	LTE-M
Voice or two-way audio	LTE-M
Firmware updates over the air on a regular basis	LTE-M, or NB-IoT with a careful update strategy
Low latency or interactive control	LTE-M
Stationary device deep indoors or underground	NB-IoT (or LTE-M with CE Mode B where available)
Very small, infrequent messages and a long battery life	Either, with PSM and eDRX tuned
Deployment across many countries on one SIM	Whichever technology your partners support in each market, often both

Availability is the deciding factor more often than the radio specs. NB-IoT and LTE-M rollouts differ by operator, and they change: AT&T stopped selling NB-IoT and planned to shut down its NB-IoT network by the end of March 2025, moving customers to LTE-M (RCR Wireless). Many module designs therefore support both and pick per market. That flexibility is useful, but it doubles the test matrix.

What to test for each technology

Attach and registration

Force each technology separately and confirm the device registers, on the expected network, within your time limit. Read the registered access technology instead of trusting the configuration. 3GPP TS 27.007 reports NB-IoT as its own access technology (“E-UTRAN (NB-S1 mode)”), and many vendors also report LTE-M distinctly. Record any attach reject cause. An NB-IoT reject on a partner that supports LTE-M usually means NB-IoT roaming isn’t enabled for your SIM range.

Coverage and coverage enhancement

Test at weak signal as well as strong. Record RSRP, RSRQ and SINR with each result, along with time to attach and success rate at the edge of coverage. For NB-IoT deployments sold on deep-indoor coverage, test in the places devices will actually be installed: basements, meter cabinets, inside equipment enclosures.

PSM behavior

Devices request Power Saving Mode timers: an active time (T3324) and an extended periodic tracking area update (T3412). The network may accept them or grant different values. The GSMA NB-IoT guide notes the maximum sleep is about 413 days and the maximum active time is 186 minutes. Test:

• Requested versus granted timers, on each home and visited network. With AT+CEREG=4, many modules report the granted active time and periodic TAU in registration status.
• Wake-up without reattach. Uplink data after sleep should flow without a full attach.
• Downlink while asleep. The guide recommends operators store at least the last 100-byte packet for a sleeping device. Check what your network actually holds and delivers on wake-up.

eDRX behavior

Request a cycle with AT+CEDRXS, read what the network granted (AT+CEDRXRDP on modules that support it), then send a downlink message and measure how long it takes to arrive. Downlink latency should track the eDRX cycle and paging window. LTE-M and NB-IoT have different maximum cycles, so the same application setting can behave differently on each.

Application round trips: MQTT and CoAP

Test the protocols your devices actually use:

• MQTT runs over TCP, usually with TLS. Measure connect time (TCP plus TLS handshake), publish-to-subscribe round trip and reconnect behavior after sleep. On NB-IoT, handshakes are a noticeable share of the energy and time budget.
• CoAP runs over UDP, often with DTLS. Measure request/response time and behavior when the device’s address binding in the operator’s NAT has expired during a long sleep.
• Keepalives versus sleep. A short MQTT keepalive defeats PSM. A long sleep can outlast NAT timeouts, which means the server can no longer reach the device until it sends uplink first.

Mobility (LTE-M)

For moving devices, test data transfer during handover and verify sessions survive cell changes. If NB-IoT is used on slow-moving or occasionally moving assets, test what happens when a device crosses cells mid-transfer. Expect re-selection and re-establishment rather than seamless handover.

Roaming support

Treat LTE-M and NB-IoT roaming as separate from 4G roaming. For each partner and technology, test attach, data, PSM/eDRX grants and, for LTE-M voice devices, VoLTE. The GSMA deployment guides recommend IP over the user plane as the minimum for LTE-M roaming and IP over the control plane for NB-IoT roaming. If devices use non-IP data delivery, test that path explicitly with each partner. See the roaming testing guide for building a partner matrix.

eSIM and SGP.32 provisioning over LPWA

If devices receive their operator profile over the air, the download itself runs over LTE-M or NB-IoT, usually on a bootstrap profile. Under SGP.32, an eIM triggers the download, and the profile package reaches the device over the cellular link, either directly from the SM-DP+ or relayed through the eIM depending on the implementation. Over NB-IoT, the secure sessions and transfer take noticeably longer and are more exposed to weak coverage than over LTE-M. Test the full download, enable and attach sequence on the technology each device will actually use, including what happens when the connection drops halfway. Also confirm that the bootstrap profile’s network supports that technology in every market. A bootstrap that only works on LTE-M can’t rescue an NB-IoT-only device. The SGP.32 testing guide and eSIM profile download errors cover the failure modes.

How to force the RAT during tests

Restricting the module to one radio access technology (RAT) makes results unambiguous. On Quectel BG95-series modules (KORE configuration guide):

Command	Purpose	Values
AT+QCFG="iotopmode",<mode>	Which LPWA technologies to use	0 = LTE-M (eMTC) only, 1 = NB-IoT only, 2 = both
AT+QCFG="nwscanseq",<seq>	Scan order	00 = automatic; 01 = GSM, 02 = eMTC, 03 = NB-IoT, concatenated in priority order
AT+QCFG="nwscanmode",<mode>	GSM vs LTE scanning	0 = automatic, 1 = GSM only, 3 = LTE only

Standard 3GPP TS 27.007 commands complete the picture:

• AT+COPS=1,2,"<MCCMNC>",<AcT> selects a specific network (PLMN) and access technology.
• AT+CEREG? confirms registration status and the access technology in use.
• AT+CPSMS and AT+CEDRXS request PSM timers and eDRX cycles.

Two practical cautions. Some configuration changes only take effect after the modem re-registers or restarts, so verify the registration result every time. And reset the module to its production configuration after testing, because a device shipped locked to one technology will fail in markets where it isn’t deployed.

An example test matrix

Device	RAT	Network	Checks
Module A	LTE-M	Home operator	Attach, data, MQTT round trip, PSM grant, eDRX grant
Module A	NB-IoT	Home operator	Same, plus deep-indoor attach
Module A	LTE-M	Roaming partner 1	Attach, data, MQTT, PSM/eDRX grant, CE mode
Module A	NB-IoT	Roaming partner 1	Attach (expect reject if not agreed), data, MQTT
Module A	LTE-M	Roaming partner 2	Same as partner 1, plus VoLTE if sold

Add a row for each partner and technology you rely on, and run the matrix again after module firmware updates, network changes or new partner agreements.

SimCheck.ai runs these checks on real modems, including the Quectel BG95-M3 on your own edge devices. Every test can force the RAT to LTE-M or NB-IoT and force a partner by PLMN. The IoT (MQTT) Quick Test attaches on the chosen technology and completes an MQTT publish/subscribe round trip, and Persistent Attach keeps a SIM attached for days for long-running behavior checks.

Test LTE-M and NB-IoT on live networks

Datasheets tell you what a technology can do. Only live-network tests show what your operators and partners actually deliver. See how IoT connectivity testing on SimCheck.ai covers attach, data and MQTT round trips on LTE-M and NB-IoT, or read how to test SGP.32 devices if your IoT fleet uses eSIM.