LTE Cat 1 vs Cat 4 vs NB-IoT

Cellular IoT • Engineering & purchasing guide

LTE Cat 1 vs Cat 4 vs NB-IoT

Start with Cat 1 for moderate data and mobile devices, Cat 4 for heavier transfers, and NB-IoT for small, infrequent messages when the power budget and command timing allow it. The right choice must also pass your operator, antenna, firmware-update and field-test requirements.

For OEM engineers, IoT product teams and component buyers. Practical calculations • Public test evidence • Module sourcing checklist

Before requesting a quote

Prepare the target countries, operator, exact order code, traffic profile and expected service life.

The short answer

Choose for the hardest required operation, not the average sensor packet. A device that reports a few bytes may still need a large firmware download, prompt remote commands or a connection that survives movement between cells.

Moderate traffic + mobility

LTE Cat 1

A starting point for trackers, connected terminals and telemetry that need conventional LTE mobility without Cat 4 throughput.

Larger transfers

LTE Cat 4

Consider it for gateways, image uploads and larger data bursts when the host, antenna system and network can use the extra capacity.

Small reports + sleep

NB-IoT

Consider it for compatible metering and sensing workloads. Verify downlink timing, updates and the actual operator service first.

Downlink (DL) means network to device, as in a firmware download. Uplink (UL) means device to network, as in an image upload. A high downlink number does not answer an uplink requirement.

Swipe the table to compare all three technologies.

Cat 1 and Cat 4 figures are maximum rates, illustrated by Quectel EC21 and EC25 documentation; they are not expected application throughput. Antenna and NB-IoT mobility distinctions are covered by u-blox and ETSI. [1] [2] [3] [4]

01 / Understand the names

Cat 1 bis is not LTE-M, and it is not a slower Cat 1

Cat 1 bis keeps the Cat 1 rate class while using one receive path. Conventional Cat 1 uses receive diversity. Removing that second path can simplify the antenna design, but it changes the receive-side trade-off. Validate the intended Cat 1 bis module in the final enclosure; do not turn a two-antenna Cat 1 design into Cat 1 bis simply by leaving a connector unused. [3]

Cat 4's downlink MIMO and Cat 1's receive diversity are not interchangeable concepts. Ask for the manufacturer's reference layout, required antenna ports and ground-clearance rules rather than specifying only “one antenna” or “two antennas.”

LTE-M commonly refers to Cat M1, not Cat 1. It belongs on a separate shortlist when low power and mobility both matter and the target operator supports it. For NB-IoT, NB1 is associated with Release 13 and NB2 with Release 14 enhancements; the label alone does not establish a usable transfer rate. [5]

A useful spec-sheet check: Quectel's BC660K-GL Hardware Design V1.0 lists 127 kbps DL / 158.5 kbps UL for its multi-tone maximum figures, and lower single-tone figures. These are that document's mode-specific limits, not universal NB-IoT speeds. [6]

02 / Start with the workload

A small daily data total can hide a difficult update

A 500-byte report every 15 minutes produces 96 reports, or 48,000 bytes of application payload per day. That is only 48 kB in decimal units. It excludes protocol overhead, acknowledgments, security handshakes and retransmissions.

Now add a 2 MB firmware package. Its delivery window may be more demanding than the normal reporting workload. Document four separate items: routine reports, alarm bursts, remote commands and maintenance transfers. Include the largest burst, not just the monthly data allowance.

Payload transfer time = payload bytes × 8 ÷ effective bits per second

Use measured application throughput in the required direction. Category peaks omit important constraints such as cell loading, radio conditions and the host interface. If the host cannot feed the modem quickly enough, a faster radio category does not remove that bottleneck.

Illustrative calculator • Not a network prediction

How long would your payload take?

Enter the transfer size and an assumed or measured effective rate. The presets explore rate sensitivity; they do not represent Cat 1, Cat 4 and NB-IoT respectively.

1 MB = 1,000,000 bytes. Enter more than 0, up to 100,000.

1 kbps = 1,000 bits/s. Enter 0.001 to 1,000,000.

Payload-only transfer time

13 min 20 sec

2 MB at 20 kbps = 800 seconds.

Not total update time. Add any connection setup, interruptions, retransmissions not already reflected in your measured rate, flash programming, verification and reboot. This result is not a battery-life estimate.

Make the update recoverable. Decide how the device resumes a partial transfer, verifies the image and recovers after a power loss. A smaller delta update can change the traffic budget, but its availability depends on your firmware architecture, not the cellular category.

03 / Separate two kinds of delay

A device that can send an alarm may still miss an incoming command

Write two different timing requirements: sensor event to server acknowledgment, and server command to device action. They are not equivalent when the modem sleeps.

In power saving mode (PSM), a device remains registered but is not listening for paging. It can wake for a local uplink event or a scheduled network update; the server cannot simply page it awake during PSM. Extended discontinuous reception (eDRX) instead provides periodic paging opportunities, trading energy for reachability. Inspect the settings actually granted by the network. [7]

PSM: no paging during deep sleep

A local alarm can wake the application and modem. That does not make the sleeping device continuously reachable from the cloud.

eDRX: periodic paging windows

The interval and paging window affect delay and current. Confirm the operator's behavior and how the server retries or queues commands.

Conceptual timing only, not to scale. PSM and eDRX can be used together. Support and usable settings depend on the module, firmware and network; they are not a promise attached to every category label.

Illustrative requirement change

The meter works — until the command deadline changes

A meter that uploads four times a day and accepts configuration at its next session can be a reasonable NB-IoT candidate. If the product now requires an unscheduled valve command within a short, fixed deadline, the previous sleep strategy may no longer fit. Revisit reachability and energy together before changing the modem.

A Cat 4 label does not create a deterministic control channel either. Safety-related actions need an appropriate local fail-safe design and a validated communication architecture, not an assumed public-network response time.

04 / Budget the whole device

Battery life is a charge budget, not an NB-IoT feature

If a product has 3,000 mAh of usable battery capacity and a five-year target, its entire average-current budget is:

3,000 mAh ÷ (5 × 365 × 24 hours)≈ 68.5 µAFor the modem, host, sensors, regulator losses and everything else combined.

“Usable” matters: available capacity depends on the battery, temperature, aging, cutoff voltage and load profile. Peak radio current can also cause a brownout even when the average-current calculation looks acceptable.

What happens when reports become four times more frequent?

For an illustrative budget, not a module measurement, assume each complete communication event uses 100 mA for five seconds: 0.139 mAh per event. Add a 10 µA background load for 24 hours, or 0.24 mAh per day.

This simplified arithmetic is not predicted service life. The full-day background term conservatively overlaps the active periods. No extra allowances are included for retries, no-service searches, self-discharge or unexpected events; 3,000 mAh is assumed already usable.

Measure charge per successful transaction at the battery input. Include wake-up, registration when needed, the application exchange and the return to sleep. Repeat with weak signal, a cold battery, failed acknowledgments, a network outage and queued reports after recovery. A tiny sleep-current figure cannot describe these events.

05 / Learn from published measurements

A public field test shows why the network belongs in the power budget

Documented engineering case • Nordic Semiconductor, 2022

The lowest-charge option changed with the test location

Martin Lesund's published field study compared LTE-M and NB-IoT using nRF9160 devices at locations approximately one kilometer apart. The devices were stationary during each test and cold-started at each location.

LTE-M used less charge at many locations, while NB-IoT had an advantage under some obstructed conditions. Network inactivity timers also differed. The study notes that power measurements and modem traces came from different devices, which limits direct correlation between the two records.

What to take from it: log the network settings and measure the whole transaction in representative locations. This was an LTE-M-versus-NB-IoT experiment, not a Cat 1-versus-Cat 4 ranking or a guaranteed range claim. Read the original field study [8].

For your own pilot, keep a record for each sample: location and enclosure, module order code, firmware, SIM/operator, radio measurements, delivered payload, transaction charge, completion time and failure reason. Keep the spread of results, not only the average. A design that works at most sites can still miss the product's service target at the difficult ones.

06 / Validate the actual deployment

Coverage, mobility and roaming are separate questions

A phone's LTE signal is not an NB-IoT acceptance test

Confirm the operator's service for the actual country and site, then test the selected module, SIM and antenna together. Indoor placement, a metal enclosure and antenna detuning can change the result.

NB-IoT coverage enhancement can use repetition to support difficult links. That can also extend delivery time and increase energy per completed message. Stronger link reach does not automatically mean better battery life. [9]

Do not replace these checks with a universal “works underground” claim. The installation is part of the radio system.

Movement between reports is not continuous connected mobility

ETSI TS 136 300 V18.3.0 lists handover among functions not supported by NB-IoT. Idle cell reselection and subsequent reconnection are different from keeping an active session through conventional LTE handover. [4]

An asset that moves, stops and then sends a short report may have a different requirement from a vehicle that must keep exchanging data along a route. For the latter, start the evaluation with Cat 1/Cat 1 bis, or Cat 4 if the traffic justifies it; also evaluate LTE-M where its capabilities and operator support fit.

Roaming is a service question, not a yes/no category slogan

Do not assume NB-IoT roaming is impossible, or that a globally advertised SIM guarantees it. Ask which destination networks are accessible, which radio technologies and features are enabled, and what restrictions apply to the deployment. GSMA's deployment guidance treats interoperability and roaming as implementation considerations. [9]

Repeat the same discipline for Cat 1 and Cat 4. Matching frequency bands is necessary, but it does not by itself establish certification, service access or the expected lifetime of a commercial network offering.

07 / Put the differences to work

Three application scenarios — and what changes the shortlist

The following are illustrative design scenarios, not reported ICNETS projects or customer results.

Scenario A / Fixed battery meter

Start with the reporting and command schedule

WorkloadSmall periodic readings, occasional alarms and a multi-year battery target.Starting shortlistNB-IoT if service is available and the sleep/reachability plan fits; compare another supported option when it does not.Decision changerA command that must arrive outside the planned receive window, or an update that takes too much energy.Acceptance testMeasure complete event charge inside the final enclosure and verify both local-alarm delivery and cloud-command timing.

Scenario B / Moving asset tracker

Test the route, not only the bench

WorkloadRegular position reports, remote configuration and changing cells while moving.Starting shortlistCat 1/Cat 1 bis for moderate traffic; include LTE-M if its power, mobility and local service are suitable.Decision changerContinuous connectivity versus store-and-forward reporting after stops. These are different service requirements.Acceptance testRoute-test session recovery, handover behavior, GNSS energy and reconnection after tunnels or no-service areas.

Scenario C / Industrial gateway

Count simultaneous traffic, not connected machines

WorkloadAggregated telemetry, diagnostic logs and occasional firmware transfers on a powered gateway.Starting shortlistCat 4 when measured aggregate traffic and transfer windows need more capacity; Cat 1 may be enough for lighter loads.Decision changerA large log upload overlapping normal telemetry and an update. If images are sent, evaluate uplink performance specifically.Acceptance testRun the combined workload through the actual host interface, antenna system, server and operator connection.

08 / Turn the shortlist into an orderable design

Buy an exact module configuration, not just “Cat 1”

A shared category or similar footprint does not make two modules drop-in replacements. Pin functions, supply peaks, interfaces, antenna connections, AT commands and firmware behavior can differ. A module approval also does not automatically satisfy every requirement for the final product.

Even within a family, features vary. Quectel's EC25 documentation, for example, describes optional VoLTE and data-only variants. If voice is required, verify the exact variant, firmware, operator provisioning and service support; do not infer voice from “LTE.” [2]

What to include in a module RFQ

For a Quectel-based shortlist, see ICNETS's Quectel sourcing page. Review the quality-assurance information and purchasing process alongside your engineering acceptance criteria.

Validate before locking the production BOM

1. Write pass/fail limits first. Define message success rate, command deadline, update time, daily charge and recovery behavior for the product's intended service.
2. Freeze the test configuration. Record module order code, firmware, SIM, APN, operator, antenna and enclosure. A development kit with a different configuration is not the final design.
3. Exercise the difficult conditions. Test weak signal, temperature limits, low battery, movement where relevant, and recovery after service loss. Include retries and queued traffic.
4. Review distributions and failures. Compare completion-time and charge distributions, unsuccessful transactions and recovery time. Run a representative pilot before accepting a production substitution.

Preparing a cellular module sourcing request?

Bring the exact order code, target markets and validation requirements into the purchasing discussion. ICNETS is a component sourcing partner; module capability and network service still need confirmation against the manufacturer's documentation and your operator.

Explore Quectel sourcingReview the purchasing process

09 / Common questions

LTE Cat 1, Cat 4 and NB-IoT FAQs

Q: Is Cat 4 better than Cat 1?

Cat 4 offers a higher maximum rate, but it is not automatically a better fit. Cat 1 may meet a moderate-data application with different hardware and cost trade-offs. Compare the required burst size, update window, power budget and total design cost on the intended network.

Q: Is LTE Cat 1 the same as LTE-M?

No. LTE-M commonly refers to Cat M1, while Cat 1 is a different LTE category. If low power and connected mobility are both important, evaluate LTE-M separately where the operator and module support it.

Q: Is Cat 1 bis slower than conventional Cat 1?

Cat 1 bis retains the same maximum rate class: 10 Mbps downlink and 5 Mbps uplink. Its main distinction is a single receive path. Actual throughput and coverage still depend on the module, antenna design and radio conditions.

Q: Can NB-IoT be used in a moving device?

Sometimes, if the application tolerates cell reselection, reconnection and the resulting delivery behavior. That is not the same as conventional connected-mode handover. Validate the route and reporting pattern; do not assume continuous-session performance from a stationary test.

Q: Does NB-IoT guarantee ten years of battery life?

No. Service life depends on usable battery capacity, reporting frequency, event charge, receive behavior, signal conditions and the rest of the electronics. Measure the complete device under representative conditions and include failure and recovery events.

Q: Can NB-IoT support firmware updates?

It can, where the module, firmware and application architecture support the required update method. Check package size, effective throughput, energy, security and recovery after interruption. A small routine telemetry payload does not prove that the firmware-update requirement is feasible.

Q: Can I use any LTE SIM with an NB-IoT module?

Do not assume so. The operator must enable the appropriate service and access for the SIM profile and destination network. Confirm the APN, bands, roaming arrangements and required features, then test the exact combination.

Q: Which option is best for a global product?

There is no category-only answer. Build a country-by-country matrix covering bands, operators, service access, certifications, roaming, required features and lifecycle plans. A regional variant or more than one validated configuration may be necessary.

Technical sources and further reading

Manufacturer documents establish product-specific features. Standards explain radio behavior; operator confirmation and field tests establish whether a particular deployment meets the application requirements.

1. Quectel — EC21 LTE Cat 1 seriesProduct example for the 10 Mbps downlink / 5 Mbps uplink maximum rate class.
2. Quectel — EC25 LTE Cat 4 seriesMaximum rates, MIMO capability and variant-dependent features including optional VoLTE.
3. u-blox — Understanding LTE Cat 1 bisSingle receive path versus conventional Cat 1 receive diversity.
4. ETSI TS 136 300 V18.3.0 — E-UTRAN overall description (PDF)Release 18 document, October 2024; section 4.10 covers NB-IoT characteristics, including mobility limitations.
5. Nordic Developer Academy — LTE-M and NB-IoTTechnology naming and NB1/NB2 release context.
6. Quectel — BC660K-GL Hardware Design V1.0 (PDF)Table 2 provides mode-specific maximum transfer figures and module interface details.
7. Nordic Semiconductor — Analysis of eDRX, PSM and AS-RAIPaging, sleep behavior and the effect of network configuration on power use.
8. Martin Lesund / Nordic Semiconductor — LTE-M vs NB-IoT field testSeptember 8, 2022. The public experiment summarized above; includes test conditions and limitations.
9. GSMA — Mobile IoT Deployment Guidelines (PDF)October 2022 guidance on deployment, interoperability and roaming. Confirm today's commercial service separately with the operator.

Photograph credits and license links appear with each image. Illustrative calculations and application scenarios are distinguished from published measurements.