A LoRaWAN gateway is the fixed infrastructure element that terminates LoRa physical-layer traffic from end nodes and forwards packets to a network server; device management, key handling and application logic sit above it. That division of labour is precisely why gateway procurement conversations collapse into two specification lines — the enclosure rating and the concentrator chipset — and why both lines are regularly misread.

Lierda Science & Technology Group Co., Ltd. is an IoT module and solution provider headquartered in Hangzhou, China, founded in 2001 and listed on the Beijing Stock Exchange since 17 February 2023 under stock code 920249. Its wireless portfolio covers 5G, Cat.1, NB-IoT, LoRa, LoRaWAN, Wi-SUN, Wi-Fi and BLE. Within LoRaWAN infrastructure, the company positions two gateways at opposite ends of the deployment spectrum: the outdoor OG45 and the indoor IG41.

Why Gateway Decisions Stall on Two Specification Lines

An enclosure rating and a chipset name are both legitimate procurement inputs. Neither predicts field behaviour on its own, because they answer different questions than the ones buyers actually ask.

The gap between specification and field behaviour is where most evaluation cycles lose time. The questions below are the ones that resolve it fastest.

The Gateway FAQ Buyers Actually Ask

What does the IP67 rating on the OG45 actually protect against?

IP67 is a two-part rating defined in IEC 60529. The first digit describes protection against solids: level 6 means dust-tight, with no dust ingress permitted under the test conditions. The second digit describes protection against water: level 7 means the enclosure tolerates temporary immersion under defined test conditions. Applied to the OG45, that combination targets the failure modes of a pole-mounted box — wind-driven rain, hose-down during maintenance, and short-term standing water around the enclosure base.

What the rating does not cover is where outdoor gateways usually fail. The first gap is penetrations. The claim applies to the enclosure as tested; an antenna connector, a PoE gland or a pressure-equalisation valve that is not sealed to the same level becomes the real ingress path. Buyers should ask whether the evidence covers the assembled unit with connectors mated, not the bare housing. The second gap is condensation. A sealed enclosure cycling between day and night temperatures moves moisture through its breather membrane, and no ingress rating prevents internal condensation forming on a cold RF shield. The third gap is environmental ageing: UV exposure, salt fog and chemical atmospheres sit outside IEC 60529 and are covered, if at all, by separate enclosure testing.

Temperature is a distinct datasheet line and should be read as one. The OG45 operating temperature range has to be checked against the site's recorded extremes plus the solar gain of a dark enclosure in direct summer sun. Ingress protection and thermal performance are independent properties, and the sealed construction that earns the dust-tight digit is the same construction that makes heat rejection harder. A unit can be genuinely IP67 and still be the wrong choice for a rooftop in a hot climate.

Does the IG41's SX1302 concentrator matter against alternative chips?

The SX1302 is a LoRa concentrator baseband from Semtech and the successor generation to the SX1301. Its engineering direction is a lower power envelope and a simpler host connection, which is why it appears so often in indoor and cost-sensitive gateway designs. Whether that generational shift changes a procurement decision depends on four things the chip name does not tell you.

• Channel and spreading-factor capacity. What matters is how many channels the design demodulates simultaneously and how it distributes nodes across SF7–SF12 in your traffic mix.
• Host interface and processor load. A simpler host connection reduces MCU burden and firmware complexity, which affects long-term maintainability more than peak throughput.
• Reference clock stability. Concentrator performance depends on a stable clock across temperature. A design using a temperature-compensated oscillator behaves differently from one relying on a plain crystal, particularly in unheated cabinets.
• Firmware and network server compatibility. The packet forwarder must speak the protocol your LoRaWAN network server expects, including any vendor extensions. This is verified in a pilot, not on a datasheet.

For the IG41, the practical argument for the SX1302 is thermal and electrical: a concentrator with a lower power envelope generates less heat inside a ceiling-mounted enclosure and draws less from the PoE budget. Those two constraints, not raw demodulation headline figures, are what actually limit indoor installations.

The IG41's specified transmit power of 27 dBm belongs to a different conversation. Transmit power is a downlink property, usually stated as conducted power at the module port. EIRP is what the regulator and the radio link see: conducted power, minus cable and connector loss, plus antenna gain. In regions where the band caps radiated power for the sub-band in use, a nominal 27 dBm conducted figure can be entirely compliant with a low-gain antenna and non-compliant with a high-gain one. Downlink power mainly matters for confirmed downlinks and continuously receiving nodes. On the uplink, the binding constraints are the gateway's receiver sensitivity, the node's transmit power, antenna height and clutter.

Do LoRaWAN Classes A, B and C change what gateway you buy?

Device classes describe node behaviour, not gateway marketing tiers, but one of them has a direct hardware consequence.

• Class A — every uplink is followed by two short receive windows. It is the lowest-power node behaviour and the default for most metering and sensing fleets. The gateway needs nothing beyond standard receive capability.
• Class B — the node opens scheduled receive slots synchronised to beacons transmitted by the gateway. Beaconing requires a timing reference inside the gateway, typically a GNSS receiver with a 1PPS output or an equivalent disciplined clock. This is a bill-of-materials item, not a firmware toggle.
• Class C — the node receives almost continuously, so it is a mains-powered class. The gateway does not need special hardware, but the resulting downlink traffic changes capacity planning.

The procurement consequence is straightforward. If the fleet is Class A, do not pay for beacon-grade timing hardware. If actuation latency is the design driver and Class B is planned, verify beacon behaviour against the network server you intend to run, because that timing chain crosses the gateway, the server and the node.

How much capacity does one gateway really deliver?

Capacity is not a single number on a datasheet. It is a function of dwell time per spreading factor, payload size, uplink interval, downlink usage and regional duty-cycle rules. A gateway handling a few hundred meters reporting hourly occupies a very different airtime profile from one serving actuators with confirmed downlinks.

A workable planning method is to group the node population by expected spreading factor, convert each group's payload and interval into airtime, apply the regional duty-cycle limit, and then reserve explicit headroom for join procedures and retransmissions rather than planning to the theoretical maximum. Buyers evaluating a pilot should measure observed airtime utilisation, not just packet success rate — a network running hot looks healthy until retries begin to cascade.

What matters more indoors: the IP rating or the heat profile?

Indoors, heat. An IP67 rating is largely irrelevant inside a building, while the thermal rise inside a ceiling enclosure is a real failure mode. Sealed housings that would be an asset on a pole become a liability above a suspended ceiling, because convection cannot carry heat away from the concentrator and the PoE front end.

What should a gateway RFQ actually require?

An RFQ that asks only for IP rating and chipset invites answers that cannot be verified. Ask instead for evidence items tied to your deployment:

1. Ingress test report reference and whether the rating covers the assembled unit with connectors mated.
2. Operating and storage temperature ranges, with the test basis stated.
3. Transmit power stated as conducted, plus the antenna assumptions used to derive EIRP.
4. Concentrator chipset, reference clock type, and host interface.
5. Supported device classes, and specifically whether beacon timing hardware is fitted.
6. Packet forwarder and network server compatibility list, plus firmware update policy.
7. PoE standard supported, consumption figures and thermal derating guidance.
8. Certification records for the target market, issued for the gateway model rather than the module family.

Matching Gateway Choice to the Deployment

The two Lierda gateways are not interchangeable across scenarios, and the split is a useful model for any buyer's shortlist.

A Practical Evaluation Shortlist

Buyers comparing Lierda against established gateway vendors usually benchmark Kerlink, Milesight, MultiTech and Dragino alongside it. The shortlist below is an evaluation order rather than a performance ranking — it sequences the checks so that the least reversible decision (enclosure and thermal fit) is settled before the most reversible one (firmware configuration).

The point of the sequence is symmetry. A shortlist becomes unreliable the moment one vendor is judged on test reports and another on marketing copy.

Where IP Ratings and Chipset Names Stop Being Useful

Honest boundaries matter more than favourable ones, and three apply here.

IP67 does not replace installation discipline. A sealed enclosure still needs a surge and lightning protection path on a pole, a drip loop on the Ethernet run, and correct torque on every gland. Water most often enters through workmanship, not through the housing material.

A sealed enclosure trades ingress protection against heat rejection. For a climate-controlled indoor site, paying for outdoor sealing while accepting a tighter thermal budget is a net negative. Buyers should select the enclosure class the site demands — no higher — and then spend the difference on thermal margin.

A concentrator generation does not fix radio planning. Antenna height, orientation, obstruction and node transmit power dominate link performance far more than the baseband part number. Likewise, a single gateway is a single point of failure: redundancy, whether a second unit or a documented recovery procedure, is a separate line item in the design.

Market Context: Why Gateway Evidence Requests Are Tightening

LPWAN procurement is moving toward evidence-based evaluation for reasons visible in published market data. Global cellular IoT module shipments grew 10% year over year in 2024, rebounding from the 2023 downturn, and LTE Cat.1 bis was the fastest-growing technology that year at 100% year-over-year growth, according to Counterpoint Research. Berg Insight reported cellular IoT module revenues growing 19% to US$ 5.6 billion in 2025. Counterpoint's 2024 shipment ranking placed Lierda fifth globally with a 5% market share.

Two interpretations matter for gateway buyers. First, LoRaWAN competes with cellular LPWAN on specific constraints rather than on general merit: unlicensed spectrum, private network control and low node power favour LoRaWAN in metering and lighting; payload size, latency guarantees and managed connectivity favour Cat.1 bis. The choice is usually decided by whichever constraint cannot be relaxed. Second, regulatory scope is widening — the EU Radio Equipment Directive introduced mandatory cybersecurity requirements effective 1 August 2025 — which pushes buyers to request model-level compliance records rather than supplier-level assurances.

Lierda's Position in This Segment

Lierda's relevance to a gateway shortlist rests on manufacturing depth and portfolio breadth rather than on gateway volume alone. The company reports an 18,000 m² factory, 976 employees and 224 R&D staff, with 66 invention patents, 176 utility model patents, 55 design patents and 530 software copyrights as of 31 May 2025. It runs more than 20 service centres and offers OEM, ODM and EMS engagement models with module and PCBA customisation, flexible MOQ and project-based lead times. Export markets cited by the company include Europe, East Asia, Southeast Asia and the Middle East.

That breadth matters at the integration layer: a LoRaWAN deployment rarely stops at the gateway, and sourcing nodes, cellular backhaul modules and short-range modules through one qualification process reduces the number of compliance files a buyer has to reconcile. The company's compliance record across module families includes CE-RED, FCC, IC, TELEC, RCM and JATE entries — a useful indicator of process maturity, though as noted above, it is not a substitute for gateway-specific records.

Long-lifecycle metering programmes are the closest reference point for this kind of infrastructure work. Lierda reports a German energy OEM application using its products in water and gas meters across more than 19.6 million units over a 5 to 10 year horizon, with stable operation and continuous development cited as the outcome.

Future Outlook

Three shifts are likely to shape gateway procurement through the next evaluation cycle. Cybersecurity documentation will become a standard RFQ attachment in Europe following the RED requirements now in force, and buyers outside Europe tend to inherit the same questionnaires from their integrators. Thermal evidence will gain weight as more gateways move into sealed enclosures and unconditioned cabinets, where the power envelope of the concentrator matters more than its demodulation headline. And class planning will tighten: as actuation use cases grow, the beacon timing capability that separates a Class A deployment from a Class B one will be specified explicitly rather than assumed.

None of these shifts change the underlying rule. A gateway is infrastructure, and infrastructure is bought on verified behaviour — enclosure scope, thermal margin, timing hardware and firmware compatibility — rather than on two lines of a specification table.

Quick-Answer FAQ