IoT geolocation: why the LoRaWAN 1% duty cycle changes everything

A team reached out to unwind a LoRaWAN trailer-tracking deployment sold on the promise that "TDoA is good enough for logistics". The problem was not accuracy: it was the duty cycle. On ETSI g1 at 868 MHz with 1% airtime, the allowed uplink rate peaks at one fix every ~55 s in practice, close to 770 m of travel at 50 km/h between two positions. At AESTECHNO Montpellier, we see this architectural mistake regularly. Updated May 2026.

IoT geolocation: the technical definition

IoT geolocation refers to the set of techniques that determine the position of a connected object, either by onboard computation (GNSS), by radio triangulation (TDoA, OTDOA, AoA), or by signal-strength estimation (RSSI). Each technology negotiates a trade-off between spatial accuracy, energy consumption and update rate, under region-specific regulatory constraints.

The market splits into two structurally different use cases: outdoor positioning for moving asset tracking (fleets, containers, trailers, livestock), and indoor positioning for high-resolution localization (warehouses, hospitals, factories). No single technology covers both. The choice is not about absolute best accuracy in the abstract, it is about fit between the technical envelope and the operational constraint.

The six dominant technologies in 2026 are multi-band GNSS (L1 + L5), Ultra Wide Band (UWB), WiFi RTT / FTM (802.11mc), Bluetooth Direction Finding (5.1 AoA/AoD), NB-IoT OTDOA (3GPP Release 14), and LoRaWAN TDoA (Semtech). Sigfox Atlas still exists but its penetration has been receding since the UnaBiz acquisition and the Atlas Native transition.

Why is the geolocation market moving in 2026?

Outdoor IoT asset tracking refers to the continuous monitoring of high-value mobile goods, from maritime containers (ISO 18185 for electronic seals) through to logistics pallets. Cold chain adds a temperature-sensing layer under EU regulation 853/2004 for perishable-goods traceability and IEC 60068-2-1 for climatic qualification of modules.

Three regulatory pressures converge in 2026. The European Cyber Resilience Act (CRA), gradually entering application since 2024, imposes security-by-design measures on every connected product placed on the EU market, which changes the calculus of radio module selection (see our comprehensive guide on CRA for IoT and embedded). The Radio Equipment Directive (RED) 2014/53/EU, revised in 2025, adds radio-cybersecurity requirements under articles 3.3(d), 3.3(e) and 3.3(f). Regulation (EU) 2023/1542 on batteries adds a carbon-footprint declaration that penalizes always-on GNSS architectures.

On the demand side, European fleet operators migrated heavily to combined cellular plus low-power GNSS modules between 2024 and 2026, after the progressive collapse of 2G/3G coverage. Sigfox saw its European base contract by more than 40% since 2022 according to UnaBiz's latest disclosures, which makes Sigfox a risky technology choice for any product targeting a 10-year commercial life.

GNSS, BLE, UWB, WiFi RTT, NB-IoT, LoRaWAN: head-to-head

A head-to-head comparison is a structured side-by-side evaluation of the six IoT geolocation technologies against uniform criteria (accuracy, update rate, consumption, indoor suitability, applicable standard), built to expose the order-of-magnitude gaps that any single datasheet figure always hides.

A head-to-head comparison of the six IoT geolocation technologies exposes a four-order-of-magnitude spread on accuracy (10 cm to 500 m), five orders on consumption (microwatts to peak watts), and two orders on practical update rate. Any single-criterion comparison is misleading.

Technology	Typical accuracy	Practical update rate	Uplink current	Indoor	Standard
GNSS L1 + L5 multi-band	under 1 m	1 to 10 Hz	20 to 50 mA during fix	no	GPS, Galileo, BeiDou, GLONASS
UWB IEEE 802.15.4z	10 to 30 cm	100 Hz and above	pulsed energy, wired anchors	yes	IEEE 802.15.4z, FiRa Consortium
WiFi RTT (FTM)	1 to 2 m	1 to 10 Hz	60 to 100 mA during scan	yes	IEEE 802.11mc
BLE AoA / AoD	1 to 5 m	1 Hz	under 10 mA	yes	Bluetooth 5.1 Direction Finding
NB-IoT OTDOA	50 to 100 m	1 per minute	120 to 250 mA during uplink	partial	3GPP Release 14, ETSI TS 138 305
LoRaWAN TDoA	100 to 500 m	1 per 1 to 2 min ceiling	20 to 130 mA during TX SF7-SF12	very partial	LoRa Alliance, Semtech

The dominant silicon vendors in 2026 are u-blox (MAX-M10S, NEO-F9P), Quectel (LC79H, L76K), MediaTek (MT3333) for GNSS, Decawave/Qorvo (DW3000) and NXP (Trimension SR150) for UWB, Nordic Semiconductor (nRF52833, nRF5340) for BLE 5.1 Direction Finding, Sequans (Monarch GM02S, GM03S) and Quectel (BC660K) for NB-IoT, and Semtech (SX1262, LR1110 with embedded GNSS) for LoRaWAN. The sourcing choice has a direct downstream impact on second-source strategy, a topic we cover in our China versus Europe electronics outsourcing analysis.

What changed in 2026

Three recent shifts weigh on the radio choice in 2026. The LoRaWAN 1.1 specification, finalized by the LoRa Alliance, generalizes secure rejoin and session-key rotation, which clears the path for the OTA updates the revised RED now requires. The LR-FHSS modulation, added through the RP002 Regional Parameters and carried by Semtech LR1121 transceivers, improves resilience in dense environments and direct satellite connectivity, a topic we develop in our NB-IoT, LTE-M and satellite connectivity comparison. Finally, the cloud-solver geolocation of the LoRa Cloud platform (the successor to LoRa Edge), paired with LR1110 modules, moves position computation off the device and onto the server, which cuts the onboard GNSS load and consumption, without lifting the ETSI duty-cycle ceiling.

Why is accuracy the wrong decision criterion?

The accuracy myth in IoT geolocation consists of choosing a radio technology solely on the headline median spatial error advertised on the datasheet, without integrating achievable update rate, target battery life, or regional regulatory constraints. This approach systematically leads to architectures that are under-spec'd in real-world operation.

The LoRaWAN case illustrates the trap perfectly. Semtech publishes a TDoA accuracy of 50 to 150 m under optimal conditions (gateway density above 4 per km², sub-microsecond time synchronization, clear urban environment). A buyer reads "150 m", thinks "good enough for trailers", and signs. The operational reality is different: a conditional accuracy of 150 m is useless if the uplink throughput is capped at one fix every two minutes, because at 90 km/h on a motorway, 120 seconds equals 3 km of travel. The position returned by the API is no longer a measurement, it is an archived record.

The mirror error exists on the GNSS side. Putting a multi-band L1+L5 receiver in a battery-powered 3-year-life sensor is technically feasible, but burns the energy budget in weeks if the receiver is always-on. The right question is never "what is the maximum accuracy", it is "what accuracy can I sustain, at the required rate, over the target lifetime, under the regulator of the target market".

The math nobody does: LoRaWAN duty cycle times SF airtime

The LoRaWAN regulatory duty cycle refers to the maximum fraction of time a Short Range Device transmitter is allowed to occupy a given sub-band, set by ETSI EN 300 220 for Europe. On sub-band g1 (868.0 to 868.6 MHz), the limit is 1%. On sub-band g3 (869.4 to 869.65 MHz), it rises to 10%. Any LoRaWAN uplink strategy must respect this ceiling, or risk fines from the national regulator and loss of CE marking.

The math nobody does before signing off on a LoRaWAN-only architecture is the following. On g1 at 1%, a device gets 36 seconds of airtime per hour. At SF7 BW125 kHz, a 10-byte payload uplink lasts roughly 46 ms (Semtech LoRa Calculator). Theoretical maximum: 36 / 0.046 = ~780 uplinks per hour, or one every 4.6 seconds. At SF10, the airtime climbs to 371 ms, which drops the cap to 97 uplinks per hour theoretical, one every 37 seconds. At SF12 (maximum range, needed to reach 3 gateways for TDoA), the airtime reaches 1.32 seconds, which caps at 27 uplinks per hour, one every 133 seconds.

And those are theoretical ceilings. In practice, you must subtract retransmissions (5 to 15% depending on QoS), downlinks (acknowledgements, ADR, MAC commands, roughly 10 to 20% of the budget), and contingency margin (jamming, interference). A robust rule of thumb: multiply the theoretical figure by 0.6 to 0.7. At SF12, that lands you at 18 to 20 usable uplinks per hour, one fix every 3 to 3.5 minutes.

For a vehicle at 90 km/h on a motorway, 200 seconds equals roughly 5 km of travel between two fixes. For a pedestrian at 5 km/h, 200 seconds equals 280 m. The verdict is unambiguous: LoRaWAN TDoA is not an outdoor solution for moving asset tracking, regardless of accuracy. It is a solution for stationary or semi-stationary assets (warehouse pallets, port-waiting containers, immobile construction equipment), with event-triggered uplinks on motion detection.

The newer LR-FHSS modulation does not change the duty-cycle arithmetic: it spreads the uplink across several frequency hops to gain sensitivity and interference resistance, but the cumulative airtime is still counted by the regulator the same way. No LoRaWAN modulation works around the ETSI 1% ceiling. The same logic applies to LPWAN alternatives: Sigfox enforces its own daily message quota, and NB-IoT shifts the constraint onto the cellular network rather than an SRD duty cycle. RSSI-based geolocation on LoRaWAN gateways, sometimes offered as a fallback when TDoA time synchronization is unavailable, degrades accuracy further (200 m to more than 1 km) without reducing the uplink budget required.

AESTECHNO methodology: how we evaluate a geolocation stack

The AESTECHNO methodology is a repeatable engineering protocol that qualifies an IoT geolocation stack before the architecture is frozen, combining use-case analysis, regional regulatory audit, RF link-budget simulation and bench measurement, so that verifiable numbers replace intuition in the radio selection.

The AESTECHNO methodology for evaluating an IoT geolocation stack runs five mandatory steps before freezing an architecture: use-case characterization (mobile versus static, indoor versus outdoor, required refresh rate), regional regulatory audit (ETSI EN 300 220 in EU, FCC Part 15 in US, KCC in Korea, ARIB in Japan), RF link budget simulation, airtime measurement on a reference bench, and battery endurance testing in a climatic chamber.

Our geolocation product specification guide mandates the explicit specification of four parameters: target accuracy measured in CEP 50 and CEP 95 (Circular Error Probable), required update rate at worst operational case, target battery life in days with specified battery capacity, and operational geography (EU, US, multi-region). Without these four numbers, the tech choice is arbitrary.

On the bench, we systematically instrument the RF chain with a Tektronix RSA306B spectrum analyzer paired with our Tektronix TekExpress for host-module validation, measure effective airtime over 24 hours of nominal operation, and characterize device duty-cycle drift against the regulatory ETSI counter. Climatic-chamber endurance (-40 to +85 °C) validates GNSS crystal and LoRaWAN PA stability over the industrial range, consistent with our EMC methodology.

Decision matrix: which technology for which operational case

The decision matrix is a two-entry selection tool that maps every combination of asset mobility and operational environment to a recommended geolocation radio stack, turning a subjective technology choice into a direct read of one row-column cell.

The decision matrix for IoT geolocation crosses two practical axes: asset mobility (static, semi-mobile, fast-mobile) and environment (outdoor, indoor, mixed). Each combination calls for a specific radio stack, sometimes hybrid. The cardinal rule: no single technology covers more than two cells out of six without a structural compromise.

Use case	Recommended stack	Why
Static container in port	LoRaWAN class A + accelerometer	No mobility, sparse uplinks, 5 to 10 year battery achievable
Trailer on motorway at 90 km/h	GNSS u-blox MAX-M10S + LTE-M / NB-IoT cellular	1 to 5 min update rate required, LoRaWAN impossible
Forklift in warehouse	UWB anchors (Qorvo DW3000) or WiFi RTT	30 cm accuracy required for navigation, wired infrastructure OK
Portable tool in hospital	BLE 5.1 AoA (Nordic nRF5340) on existing infra	1 to 5 m accuracy sufficient, 1 to 3 month battery acceptable
Fixed agricultural sensor in open field	LoRaWAN class A with embedded geolocation (Semtech LR1110)	Position computed once per day, 8 to 10 year battery life
Urban shared bicycle	GNSS L1+L5 + LTE-M + accelerometer wake-up	Mixed indoor/outdoor, theft-detectable, cellular cost only on movement

For hybrid multi-technology architectures, compare integrated modules (GNSS receiver embedded in a LoRaWAN transceiver such as Semtech LR1110, or BLE + WiFi RTT stack such as ESP32-C6) with bare-metal separately-sourced approaches. The integrated path shortens time-to-market but locks the BOM to a single supplier, which becomes critical under shortage. See our electronic component shortages and mitigation strategies analysis for the BOM framework.

Why did we unwind a LoRaWAN-only deployment?

On a recent project for cold-chain trailer tracking, in our AESTECHNO lab in Montpellier we reproduced the ETSI EN 300 220 math on a pilot fleet equipped with LoRaWAN only. The original promise: "TDoA is good enough for logistics accuracy, 5-year battery life". The measurement: at SF10 forced by real-world range (not the marketing range), the device sustained 18 to 20 usable uplinks per hour, one fix every 3 to 3.5 minutes. For a trailer at 80 km/h, that is 4.4 km of travel between two reported positions. Our measurement methodology stays consistent on every geolocation project: airtime characterization on a Tektronix RSA306B bench coupled with our Tektronix TekExpress for host-module validation, device duty-cycle drift measured against the ETSI spec over 24 hours, climatic-chamber battery endurance, all referenced against our comprehensive LPWAN technology comparison and LoRa Alliance documentation. Contrary to the common assumption that dropping GPS accuracy fixes the problem, the field report is unambiguous: accuracy is not the limiting factor, the regulator-imposed uplink throughput is. The integration team's field report confirmed this. In our practice across outdoor asset-tracking engagements, we have observed that the hybrid GNSS + LoRaWAN architecture, with uplinks triggered by accelerometer events or geofence crossings, systematically outperforms both pure approaches. Despite the added complexity of a low-power GNSS receiver (u-blox MAX-M10S or Quectel L76K), we recommend this architecture for any mobile asset above 5 km/h average speed.

Smart hybrids: GNSS plus LoRaWAN with event-triggered uplinks

The smart hybrid GNSS + LoRaWAN architecture combines a low-power GNSS receiver for spatial accuracy with a LoRaWAN radio for position backhaul, orchestrated by a low-power microcontroller (STM32WLE5, nRF9160) that only triggers uplinks on a business event (motion detected by accelerometer, geofence crossing, temperature alert). This pattern resolves both the duty-cycle constraint and the GNSS consumption issue simultaneously.

In practical terms, the firmware integrates three sub-systems: an always-on ultra-low-power accelerometer (LIS2DH12 at ~2 μA), a GNSS receiver woken on event only (u-blox MAX-M10S with offline assist or MGA Online, 2 to 5 second fix), and a LoRaWAN class A transceiver configured at SF7 or SF8 depending on coverage, transmitting positions every N minutes minimum + on event. The result on the AESTECHNO bench: 4 to 6 year battery life on a 17 Ah lithium thionyl cell, with 60 to 120 geolocated fixes per day depending on usage profile.

The secret is not in the raw performance of each sub-system, it is in rigorous event-driven scheduling and fine sleep-mode management. Our embedded power management strategy for long-life batteries details the firmware patterns tested on our Nordic PPK2 (Power Profiler Kit 2) and Otii Arc benches, which measure device consumption to the microsecond.

CE/RED and FCC certification: what changes per technology

CE/RED and FCC certification is the regulatory conformity procedure that authorizes a radio product to be placed on a market, on the European Union market under the RED 2014/53/EU directive and on the United States market under FCC Part 15, each governed by its own harmonized standards.

CE/RED and FCC certification of a geolocated IoT product varies significantly with radio technology choice: CE marking under RED 2014/53/EU covers every radio technology but with different harmonized standards (EN 300 220 for LoRaWAN, EN 300 328 for BLE and 2.4 GHz WiFi, EN 301 511 for GSM/UMTS, EN 301 908 for LTE/NB-IoT), whereas the US FCC Part 15 distinguishes subpart B (unintentional radiators) from subpart C (intentional radiators).

For a combined GNSS + LoRaWAN + BLE geolocation module, the RED technical file typically requires three distinct test reports (EN 300 220 for LoRaWAN at 868 MHz, EN 300 328 for BLE at 2.4 GHz, and an EMC EN 301 489-3 emissions/immunity test) plus the signed EU Declaration of Conformity. A full accredited-laboratory file fluctuates with product complexity. For our complete CE/RED certification guide for IoT products, the detailed path covers realistic timelines and pitfalls to avoid.

On the cybersecurity side, the revised RED 2025 tightens articles 3.3(d), 3.3(e) and 3.3(f) with mandatory secure OTA updates and strong identifier management, which now excludes geolocation firmware without signed-update support. See our industrial IoT cybersecurity framework for applicable signing and secure-boot patterns. LoRaWAN key provisioning (DevEUI, AppKey, the 1.1 session keys) and protecting the reported position against GNSS spoofing belong to the same chain of trust, detailed in our guide to securing an IoT product from design to deployment.

Bottom line

Modern IoT geolocation is no longer about datasheet spatial accuracy, it is about combining accuracy, achievable update rate under regulatory constraint, target battery life, and operational environment. The LoRaWAN duty-cycle calculation under ETSI EN 300 220 is the tool that separates realistic architectures from marketing promises.

• The ETSI 1% duty cycle caps LoRaWAN long before accuracy does. At practical SF10, 60 to 70 usable uplinks per hour. At SF12, 18 to 20 usable fixes per hour, one every 3 to 3.5 minutes.
• LoRaWAN TDoA is not an outdoor solution for moving assets. 4 km of travel at 80 km/h between two SF12 fixes. Reserved for stationary or semi-stationary assets.
• Hybrid GNSS + LoRaWAN with event-triggered uplinks dominates mid-mobility outdoor. 4 to 6 years of battery life on 17 Ah lithium, 60 to 120 fixes per day depending on profile.
• UWB and WiFi RTT dominate high-precision indoor. 10 to 30 cm with UWB, 1 to 2 m with WiFi RTT. Wired infrastructure or anchors required.
• The revised RED 2025 excludes geolocation firmware without OTA signing. Plan the secure-boot chain from the specification phase.

FAQ: IoT geolocation, duty cycle, certification

What LoRaWAN duty cycle is allowed in Europe?

The LoRaWAN duty cycle in Europe is set by ETSI EN 300 220 depending on the sub-band used. On sub-band g1 (868.0 to 868.6 MHz), the limit is 1%, or 36 seconds of airtime per hour. On g3 (869.4 to 869.65 MHz), it rises to 10%. In the United States, FCC Part 15.247 imposes no duty cycle but limits dwell time to 400 ms per channel under frequency hopping, which entirely reshapes the uplink strategy. Every LoRaWAN architecture must explicitly model this ceiling before freezing the product.

Why is multi-band GNSS L1+L5 displacing L1-only in 2026?

Multi-band GNSS L1+L5 corrects in real time the ionospheric error, the main noise source in single-band reception. The u-blox MAX-M10S, NEO-F9P and MediaTek MT3333 multi-band chipsets now reach sub-metre accuracy within seconds of TTFF (Time To First Fix), compared with the typical 5 to 10 m of L1-only. The consumption gap has shrunk to roughly 30% additional on recent modules, which makes L1+L5 the default choice for any new asset-tracking product with a commercial life above 5 years.

When is UWB right over BLE 5.1 AoA indoor?

UWB IEEE 802.15.4z is the right pick when target accuracy is below 1 m (typically 10 to 30 cm), positioning latency must stay under 100 ms, and a wired anchor infrastructure is acceptable (Qorvo DW3000 or NXP SR150 with PoE). BLE 5.1 AoA remains preferable when 1 to 5 m accuracy is enough, BLE infrastructure already exists (badges, hospital or office beacons), and tag battery life above 6 months matters. Rule of thumb: UWB for AGV and robotic navigation, BLE AoA for tracking people or portable tools.

NB-IoT or LTE-M for fleet tracking?

NB-IoT and LTE-M serve different cases. LTE-M (Cat-M1) supports mobility with handover, reaches usable throughput of 300 kbps to 1 Mbps, and allows second-by-second update rates with no duty-cycle constraint. This is the choice for moving-vehicle tracking. NB-IoT (Cat-NB1/NB2) does not support handover in class 1 (3GPP Release 13/14 limitation), which makes fixes unreliable for any mobile above 30 km/h. NB-IoT is in turn superior in indoor coverage and low power, so it fits fixed meters, parking sensors, port beacons. For motorway fleet tracking, LTE-M is the non-negotiable choice in 2026.

Does CRA 2024 impose anything specific for geolocation?

The Cyber Resilience Act regulation (EU) 2024/2847 imposes security-by-design requirements on every connected digital product placed on the EU market, including secure signed OTA updates, strong identifier management, and a minimum 5-year support life after last marketing date. For IoT geolocation, this concretely means any firmware integrating a GNSS receiver and a LoRaWAN module must be able to receive signed updates (MCUboot, RAUC, or equivalent), keep a security event log, and expose an up-to-date CBOM (Cryptography Bill of Materials). Non-compliant products will see their CE marking blocked once the regulation reaches full application.

How to estimate consumption of a GNSS + LoRaWAN tracker?

Estimating total consumption of a GNSS + LoRaWAN tracker before prototype runs through a six-line spreadsheet: sleep current of MCU + always-on sensors (typically 5 to 20 μA), average GNSS current weighted by fix-to-sleep ratio (10 to 30 mA during a 2 to 5 s fix), average LoRaWAN current weighted by airtime (20 to 130 mA during uplinks of 50 ms to 1.3 s depending on SF), leakage current of crystals and LDOs (1 to 5 μA), battery self-discharge over target life (typically 1 to 3% per year for LiSOCl2), and a safety margin (usually 30%). Our Nordic PPK2 bench then validates this estimate to the microsecond. Without this step, real-world battery life diverges by a factor of 2 to 4 from a naive estimate.

Related articles

• LPWAN, LoRaWAN, NB-IoT, Sigfox: complete comparison: the LPWAN baseline before tackling geolocation.
• Bluetooth 5.4 PAwR for IoT sensor architectures: the most recent BLE radio spec for low-power deployments.
• CE/RED certification for IoT products: the regulatory path for combined radio modules.
• Embedded power management for long-life batteries: firmware patterns that make the battery-life difference.
• Cyber Resilience Act: IoT and embedded compliance guide: what CRA changes for tracker secure-boot chains.