NB-IoT vs LTE-M for cellular IoT devices

When a device sits somewhere with no Wi-Fi and no LoRaWAN gateway you can install, cellular is the remaining option, and the two LPWAN technologies designed for small devices are NB-IoT and LTE-M. They come from the same standards body, run on the same networks, and are often sold on the same module. They still make different trade-offs, and the choice tends to be decided by two questions: does the device move, and how much does it need to say.

What NB-IoT and LTE-M are

Both are part of the LTE family of standards, defined by 3GPP, the same body that defines the mobile standards phones use (both arrived in its Release 13), and both were created to serve devices that need little data, long battery life and cheap modules. Carriers run them on their existing LTE networks, often in a narrow slice of spectrum, and both are carried forward into the 5G era so devices built on them have a long life ahead.

NB-IoT, narrowband IoT, uses a very narrow channel. That narrowness is what gives it its strengths: the signal is concentrated, so it reaches further into buildings and underground, and the modules are simple and cheap. It is designed for devices that send small amounts of data infrequently and stay in one place.

LTE-M, sometimes written as LTE Cat-M1, uses a wider channel. It supports more bandwidth, lower latency, movement between cells while connected, and voice. Modules are a little more complex. It is designed for devices that need something closer to a real connection, including ones that move.

Both offer the same two power-saving mechanisms, described below, which is why either can run a battery device for years.

Bandwidth and latency

LTE-M carries far more data than NB-IoT and responds faster. It is comfortable with firmware updates of a reasonable size, with sending a photo now and then, with a device that reports every few seconds, and with an application that needs an answer back within a second or so. It supports a full IP stack in the ordinary way, so standard protocols such as MQTT over TLS work as they do on any network.

NB-IoT is built for small, infrequent messages. Its data rates are low, its latency can be seconds, and a large transfer such as a firmware image is slow enough that designers plan carefully around it, splitting updates into small pieces or scheduling them for quiet periods. The protocol stack on top is often trimmed as well: lightweight transports and compact message formats are the norm (CoAP over UDP is the usual choice), because every byte is expensive in time and power.

For a meter, a level sensor or a simple tracker that reports position occasionally, NB-IoT’s limits are irrelevant. For a device with a screen, a sensor that streams, or anything that must respond promptly to commands, LTE-M is the one that behaves like a network.

Mobility and handover

This is the clearest dividing line. LTE-M supports handover: a device moving from one cell to the next keeps its connection, as a phone does. That makes it suitable for vehicles, trailers, containers, pallets and wearables, and for anything that moves during a session.

NB-IoT does not support connected-mode handover in the same way. A device that moves has to reconnect when it enters a new cell, which is acceptable for something that moves occasionally between reports, such as a bin or a piece of equipment, and unsuitable for anything that needs to stay connected while in motion.

A related consideration is roaming. Both technologies are widely deployed, but roaming agreements between carriers took longer to arrive for NB-IoT, and coverage maps and roaming support differ by country and carrier. A device that will be sold or moved across borders needs the carrier’s confirmation for the specific technology in the specific countries, and a SIM whose roaming profile covers them, and LTE-M has generally been the safer bet for international devices.

Power saving: PSM and eDRX

Both technologies share two mechanisms that make battery operation possible, and understanding them explains most of the battery life claims.

Power saving mode lets a device tell the network it is going to sleep for a long period, from minutes to days, and stay registered while it does. The radio is off, the network does not expect to hear from it, and when the timer expires the device wakes, sends, and goes back to sleep without having to reattach to the network. That reattachment is the expensive part of cellular, so avoiding it is where the savings come from. While the device is in this mode it cannot be reached; commands wait until it wakes.

Extended discontinuous reception lets a device that wants to remain reachable listen for the network only at agreed intervals, from seconds to some minutes, rather than continuously. Between listens the receiver is off. It uses more power than the sleep mode but the device can be contacted within the interval.

A device design chooses between them, or combines them: sleep deeply between scheduled reports, and use the listening interval during a window when commands are expected. The carrier has to support the timers the device asks for, which is worth checking early; not every network honors every value.

Coverage and module choice

NB-IoT’s narrow channel gives it better penetration: it reaches further into basements, meter cupboards and underground chambers, and further from a tower in open country. LTE-M also improves on ordinary LTE in this respect, through repetition of transmissions, but NB-IoT goes deeper. For a device installed somewhere hard to reach by radio, that difference can decide whether it connects at all.

Carrier deployment is the other half of coverage. Some carriers deployed one technology first and the other later or not at all, and the picture differs by country. A device design should start with the coverage maps of the carriers in its target markets for the specific technology, and end with a test device in the actual installation location, because a coverage map cannot see the inside of a concrete stairwell.

Modules complicate the choice in a helpful way: most cellular IoT modules sold today support both NB-IoT and LTE-M, and many add a fallback to older 2G networks where those still exist. The firmware picks a technology based on what is available and what the device is configured to prefer. That lets a product ship with one hardware design and choose per deployment, at a slightly higher module cost than a single-mode part. It does not remove the need to decide, because the device’s application design (message sizes, expected latency, mobility) still has to match the technology it will mostly use.

Which one to pick

Pick NB-IoT when the device is fixed, sends little, and is installed somewhere radio finds hard to reach. Pick LTE-M when the device moves, needs to be commanded promptly, needs firmware updates of any size without heroics, or needs anything resembling a real connection. Pick a dual-mode module when the product will be deployed in more than one country or more than one kind of location, and configure a preference per deployment. If the device needs far more data than LTE-M carries, such as video or a router’s traffic, look at 5G RedCap, the reduced-capability class of 5G built for that middle ground.

In every case, test with the real module in the real location before committing to a design, and confirm the carrier supports the sleep timers the power budget depends on. Cellular IoT works well; it works best for people who checked those two things before ordering a thousand devices.