LoRaWAN vs NB-IoT vs Sigfox: Budget an IoT Pilot

An LPWAN pilot is not priced by its radio module alone. Coverage work, enclosures, installation, certification, firmware, network service, backhaul, cloud integration, monitoring, field replacement, security, and support can outweigh the connectivity line. A private network can avoid an operator subscription while introducing infrastructure and operating responsibilities of its own.

Use current written quotes and test the exact sites, payloads, transmission pattern, device class, power source, and recovery behavior. LoRaWAN, NB-IoT, and Sigfox 0G each cover multiple commercial and deployment arrangements. The protocol name does not establish price, coverage, service level, or battery life.

What are you actually comparing?

• LoRaWAN. The standard supports public, shared, private, community, and enterprise deployments. It does not prescribe one commercial model. A private deployment normally includes gateways, IP backhaul, a network server, keys, monitoring, updates, spares, and an operator. The LoRa Alliance explains these deployment freedoms in its LoRaWAN developer overview.
• NB-IoT. This is a 3GPP-standardized cellular LPWA technology deployed by mobile operators in licensed spectrum. Band support, roaming, power-saving features, coverage, data routing, certification, and commercial terms vary. The GSMA's current Mobile IoT deployment guidance highlights why operator configuration affects performance, roaming, cost, and battery behavior.
• Sigfox 0G. UnaBiz owns the technology and works with national operators. Coverage and contracts remain region-specific. Sigfox documentation limits uplink payloads to 12 bytes and downlink payloads to 8 bytes, but usable message frequency also depends on radio configuration, regulation, and contract. Check the official payload documentation and local operator terms.

Build the budget from complete workstreams

Request comparable quotes using the same device count, countries, data pattern, retention, support hours, and contract duration. Separate one-time engineering from reusable product work and recurring service. Record taxes, currency, minimum commitments, volume tiers, overage, activation, roaming, and renewal assumptions.

Use a transparent TCO model

A useful model exposes inputs instead of claiming a universal crossover:

Total cost = engineering + device hardware + installation + network infrastructure + connectivity + cloud services + operations + replacements + decommissioning.

For each year, model active devices, failure rate, messages, data volume, site visits, batteries, support, gateway or backhaul replacement, and contract changes. Run at least base, low-density, high-density, and poor-coverage scenarios. Discounting may matter for long programs, but uncertainty ranges are usually more useful than false precision during a pilot.

A private LoRaWAN gateway does not make later devices free. Device hardware, provisioning, airtime capacity, network-server or hosting costs, security, monitoring, backhaul, maintenance, and support remain. Operator connectivity also is not necessarily a simple annual per-device fee. Pricing may include bundles, minimums, roaming, traffic classes, or managed-service components.

Test coverage instead of copying range figures

Published kilometer ranges are not site evidence. Building materials, terrain, antenna orientation, interference, frequency plan, transmit power, gateway height, network configuration, operator deployment, and regulatory limits all affect the link. LoRaWAN regional parameters define different channels, data rates, power, dwell-time, and related constraints by region. Do not reuse a result across countries without checking the applicable plan.

Measure representative indoor, basement, rural, moving, and edge locations. Record signal metrics, delivery success, latency, retransmissions, downlink success, and recovery after an outage. For private LoRaWAN, test gateway and backhaul failure. For operator networks, confirm the exact provider, supported bands, roaming behavior, and service availability at every target site. The GSMA maintains a current Mobile IoT deployment map, but the map is a screening tool, not a site guarantee.

Calculate battery life from the workload

Battery life depends on cell chemistry and capacity, temperature, self-discharge, sleep current, sensor warm-up, processing, payload size, transmission interval, repetitions, coverage level, retries, downlinks, firmware behavior, and the required end-of-life margin. NB-IoT power-saving mode and extended discontinuous reception settings also vary across networks. Poor coverage can increase repetitions and energy use.

Build an energy budget from measured current profiles on the target hardware, then validate it over representative temperature and signal conditions. Treat a ten-year headline as a use-case-specific design target, not a property guaranteed by any of the three radio names.

How should Sigfox 0G be evaluated now?

Sigfox is not accurately described as dead. UnaBiz acquired Sigfox SA and Sigfox France SAS in 2022 and currently states that it owns Sigfox 0G technology and collaborates with national network operators. See UnaBiz's acquisition notice and current service overview.

That corporate status does not prove coverage, continuity, support, or suitability for a particular country. Verify the local operator, coverage, contract term, message plan, downlink behavior, roaming, backend access, export path, and exit options. The familiar 140-uplink daily figure applies to documented radio and contractual conditions, including the European RC1 example, not every global deployment. Use the local radio configuration and contract as the authority.

Can you switch networks later?

A single-radio device normally cannot change to an incompatible radio technology through software alone. Some products use multi-radio hardware, replaceable communications modules, or a separate modem boundary, but these add cost, size, certification, firmware, antenna, and power trade-offs. Even when the radio stays the same, changing an operator or network server can require provisioning, keys, backend changes, or field access.

Reduce lock-in by separating application data from transport, documenting codecs and device identity, retaining export rights, automating provisioning, protecting key ownership, and testing recovery. Decide which components can be replaced remotely and which require a truck roll before scaling the fleet.

What should the pilot prove?

1. Coverage: the required delivery and downlink behavior at representative sites and edge conditions.
2. Power: measured current profiles and a defensible lifetime model with environmental margin.
3. Capacity: payload, reporting interval, bursts, retries, downlink, duty-cycle, and fleet-growth assumptions.
4. Operations: provisioning, monitoring, outage recovery, updates, support, replacement, and access control.
5. Economics: comparable written quotes and scenario-based total cost over the expected service life.
6. Exit: data export, keys, device ownership, contract termination, migration boundaries, and decommissioning.

Once the network choice is narrowed, run captured packets and vendor JavaScript through our free LoRaWAN payload codec tester, then keep the fixtures as regression tests.

"Choose an LPWAN from measured sites, power profiles, current contracts, and an explicit operating model. The radio label alone does not price or de-risk the pilot."