The Word “5G” Is Doing Too Much Work

There is a conversation I have had dozens of times over the past three years, in various forms. An engineer or IT manager calls because their 5G router is showing LTE. Or their IoT SIM will not connect above Cat-4 speeds. Or their installation is pulling 8 Mbps while the phone next to it is doing 400 Mbps on “5G”.

The frustration is understandable. The marketing has been clear: 5G is the new thing, 5G is fast, buy 5G equipment. What the marketing has not been clear about – and what the industry has done a poor job of explaining to the people who actually deploy this technology – is that “5G” is not one thing. It is a family of technologies, standards, device categories, and network architectures that share a name and very little else.

Understanding the distinctions does not require a radio engineering degree. It requires understanding three things: what network architecture your SIM and device are connecting to, what 3GPP device category your hardware actually supports, and whether your antenna is working with your hardware or against it. Get all three right and cellular IoT works the way it should. Get any one of them wrong and you will be troubleshooting connections for longer than any project budget allows.

I am going to walk through all three in the depth they deserve.

5G NSA vs 5G SA: The Distinction That Explains Most Connectivity Problems

When operators started deploying 5G from 2019 onwards, they faced a commercial reality: building an entirely new network from scratch takes time and costs billions. The faster path to market was to deploy 5G radio access (the masts and antennas) while keeping the existing 4G core network running the back-end intelligence. This is called 5G Non-Standalone, or NSA.

In NSA mode, a 5G NR (New Radio) base station – called a gNB – operates as a secondary node alongside an existing 4G LTE base station, called an eNB. The 4G base station remains the master node, handling signalling and control. The device connects to both simultaneously in a dual connectivity configuration. The 5G radio layer adds bandwidth and raw throughput. But the core network – the system that handles authentication, session management, policy enforcement, and roaming – is still the 4G Evolved Packet Core (EPC). From the network’s perspective, a 5G NSA device is behaving like an LTE device with a faster radio.

5G Standalone (SA) is architecturally different in every meaningful sense. In SA, both the radio access and the core network are fully 5G. The device connects directly to a 5G gNB which communicates with the 5G Core (5GC). There is no 4G anchor. Registration, authentication, and all session management use native 5G procedures. The intelligence of the network is genuinely new architecture – Service Based Architecture (SBA), network slicing, ultra-low latency support – not LTE with a faster radio bolted on.

Why this matters for your IoT SIM

Here is the scenario that causes the most confusion in the field. You install a 5G router. The customer’s phone, sitting next to the router, shows a 5G icon and pulls 400-500 Mbps. Your router is showing LTE, or at best a weak 5G connection struggling above 20 Mbps.

There are several things that could be happening – and understanding NSA vs SA explains most of them.

First: most multi-network IoT SIMs and roaming SIMs are provisioned for LTE roaming agreements. The commercial agreements between IoT MVNO operators and host networks are almost always based on LTE/4G MVNO relationships. When a 5G NSA network sees a roaming IoT SIM, it may authenticate and route it through the LTE core regardless of what the radio environment looks like, because that is the agreement in place. The phone next to it is on a consumer postpaid contract with its home operator, which has explicit 5G service tier agreements. Same mast, completely different network treatment.

Second: the NSA requirement for a 4G LTE anchor means that if the 4G signal at your installation location is weak or congested, 5G NSA will not work well even if the 5G signal itself is strong. The 5G radio needs a functioning LTE control channel to operate. This is one of the less-documented failure modes of NSA deployments and one that catches out engineers who focus on 5G signal strength without checking the underlying LTE quality.

Third: on 5G SA networks, roaming IoT SIMs face a different challenge – the SIM’s home network needs to have 5G SA roaming agreements with the visited network. These are newer, less ubiquitous agreements than LTE roaming. A multi-network SIM that cheerfully roams across all four UK operators on LTE may not have SA roaming in place, resulting in the device falling back to LTE even when SA coverage exists.

How to check whether you are on NSA or SA

On most Android devices, field test mode or a third-party app such as Network Cell Info or SYKIK can show the network registration type – look for NR SA versus NR NSA in the connection details. On iOS, Apple does not expose this information in field test mode in a user-friendly way, though some diagnostic apps get closer. For routers, Teltonika’s RutOS and similar platforms typically show the connection mode in the mobile status interface – look for NR SA, NR NSA, or simply LTE as the reported connection type.

The practical test that often reveals more than app diagnostics: lock the router to 5G only (removing LTE as a fallback option in band selection). On a 5G NSA network without a strong LTE anchor, the connection will drop or become unusable. On a 5G SA network, a SA-capable device will maintain connection. This tells you whether the SA infrastructure is actually available at your location independently of what the device happens to be negotiating by default.

LTE Categories and 5G Device Classes: What Your Hardware Actually Supports

The second layer of confusion sits in the hardware itself. “5G router” is a marketing category, not a technical specification. A router sold as “5G” may support 5G NSA only. It may support Sub-6 GHz 5G but not mmWave. It may have a module that supports 5G SA in theory but lacks the software to negotiate SA registration on a specific operator’s network. And its fallback LTE capability – which will determine its real-world performance in most European deployments for the next several years – may be Cat-4, Cat-6, or Cat-12, with dramatically different throughput implications.

Understanding the LTE category system is essential for anyone specifying cellular connectivity equipment. The categories are 3GPP-standardised device classifications that define maximum downlink and uplink throughput based on the number of MIMO layers, modulation schemes, and carrier aggregation capabilities supported by the device.

The Cat-4 ceiling and why it matters

Cat-4 has been the backbone of industrial IoT connectivity in the UK and Europe for over a decade. 150 Mbps theoretical downlink, 50 Mbps uplink, 2×2 MIMO, no carrier aggregation. In real-world conditions – good urban site, decent signal, not congested – you might see 40-80 Mbps down, 15-30 Mbps up. That is entirely adequate for the vast majority of industrial and commercial applications: remote monitoring, broadband failover, CCTV uplink, SCADA connectivity, EV charging management.

The problem is not the performance ceiling of Cat-4 – for most applications it is not a ceiling at all. The problem is the network trajectory. Cat-4 devices operate on LTE. LTE is a technology that operators are deprioritising investment in as they build out 5G. It will not disappear next year or the year after. But a device deployed today on Cat-4 with a five-to-ten year operational life expectation is being deployed on infrastructure that will progressively receive less investment, less spectrum allocation, and eventually face retirement. That is a planning risk that Cat-4 deployments made in 2020 could reasonably ignore; Cat-4 deployments made in 2026 arguably cannot.

The other Cat-4 limitation is that it is specifically a ceiling for 5G RedCap IoT SIM connections. A 5G RedCap router – the next generation of industrial cellular router hardware – connecting on a roaming IoT SIM without 5G SA roaming provisioned will fall back to LTE. Its fallback capability is typically Cat-4 or Cat-6 LTE. So you can have brand new RedCap-capable hardware and still be running at Cat-4 throughput because the SIM and the network have not caught up with the device. This is the exact scenario I described at the top of this piece – and understanding why it happens is the first step to resolving it.

A note on 5G RedCap specifically

RedCap – 5G Reduced Capability, standardised in 3GPP Release 17 – is the device category that sits between LTE mid-tier and full 5G NR. It delivers LTE Cat-4 equivalent throughput with improved latency, operates natively on 5G SA networks, and benefits from everything SA brings: network slicing, enhanced positioning, 5G security architecture, and a network technology that operators are actively investing in for the long term.

For industrial IoT deployments – routers, gateways, surveillance infrastructure, fleet terminals – RedCap is the logical succession plan from Cat-4 and Cat-6. The ecosystem is maturing rapidly. Module vendors including Telit Cinterion, Quectel, and Fibocom have commercial RedCap modules in production. The first RedCap industrial routers are shipping. European operator SA buildouts are progressing. The deployment window in Europe is 2026-2028 for early commercial rollout, 2027-2029 for volume. If you are specifying hardware with a 7-10 year operational life today, it is worth knowing whether RedCap capability is available in the product line you are evaluating.

The Antenna Problem: Why the Wrong Antenna Wastes Good Hardware

In my experience, antenna selection is the most consistently underestimated element of cellular IoT deployments. Engineers spend significant time selecting the right router, evaluating SIM providers, and configuring network settings – and then fit a cheap omnidirectional antenna because it came with the router or because the one that fits the connector was in the parts bin.

The antenna is not a passive component. It is the physical interface between your device and the network. Its gain, radiation pattern, frequency coverage, VSWR, and physical placement determine the quality of the RF path between your device and the serving cell. A router with a marginal signal and a well-matched, appropriately-sited antenna will outperform the same router with a good signal and a poorly matched or incorrectly placed antenna. This is not theoretical – it is observable in the field every time.

Frequency bands: match the antenna to the network

LTE and 5G operate across a wide range of frequency bands. In the UK and Europe, the bands you most commonly encounter are:

The practical implication of this band landscape is that a genuine multi-band antenna for LTE and 5G needs to cover 700 MHz through 3.8 GHz – a 5:1 frequency ratio. Cheap antennas claiming this coverage typically do so with significant gain variation across the band, performing well at one frequency and poorly at others. A Cat-4 router with a 5G antenna that performs poorly at 800 MHz will struggle in any location where Band 20 is the serving or anchor band – which in rural and suburban UK means most locations.

MIMO and antenna count: the configuration that actually matters

MIMO (Multiple Input Multiple Output) is the technology that allows cellular radios to transmit and receive multiple data streams simultaneously on the same frequency, increasing throughput without requiring additional spectrum. LTE Cat-4 uses 2×2 MIMO downlink – two transmit antennas at the base station, two receive antennas at the device. Cat-12 and above use 4×4 MIMO. 5G NR devices typically support 4×4 or higher.

The critical point for hardware specification and antenna selection: MIMO only works if the two (or four) receive paths are sufficiently isolated from each other. If you connect two antennas that are poorly isolated – either because they are physically too close together, or because the antenna design does not provide adequate cross-polarisation isolation – the MIMO spatial streams interfere with each other and the throughput benefit is lost. You can end up with a Cat-6 router, two antennas, and Cat-4 equivalent performance because the antenna configuration is not allowing MIMO to function correctly.

For a Cat-4 device: two antenna connections, minimum 10 dB isolation between ports, antennas ideally separated by at least half a wavelength at the operating frequency (around 20 cm at 800 MHz, around 4 cm at 3.5 GHz), or using cross-polarisation for compact deployments.

For a Cat-6 or 5G device with 4×4 MIMO: four antenna connections, same isolation requirement per port pair, physical separation or cross-polarisation between all four elements. This is where compact omnidirectional antennas often fall short – maintaining adequate isolation between four paths in a single housing is genuinely difficult.

External vs embedded antennas: when to use which

Coaxial cable: the gain you give back

Every metre of coaxial cable between an antenna and a router introduces insertion loss. At 800 MHz, standard RG58 loses approximately 0.3 dB per metre. At 3.5 GHz – the primary 5G band – that loss rises to approximately 0.9 dB per metre. A 5 metre run of RG58 to a 5G antenna costs you 4.5 dB of signal at n78 frequencies. A 3 dBi antenna 5 metres away on RG58 at 3.5 GHz is delivering a net negative gain at the device compared to no antenna at all.

The fix is either shorter cable runs, higher quality low-loss cable (LMR-195 equivalent or better for runs up to 5 metres at 5G frequencies; LMR-240 equivalent for longer runs), or positioning the router physically closer to the antenna rather than running cable from a wall-mounted antenna to a rack-mounted device. This is basic RF engineering but it is routinely ignored in installations where cellular connectivity is treated as an afterthought rather than a designed element of the system.

Putting It Together: A Diagnostic Framework for Underperforming Deployments

When a cellular IoT deployment is not performing as expected – the router is on LTE when it should be on 5G, throughput is below what the hardware should support, the connection is unstable – there is a systematic order of investigation that saves time and avoids the mistake of replacing hardware that is not the problem.

Step 1: Establish what the network actually looks like at the location

Use a phone with a network diagnostics app (Network Cell Info on Android, or simply checking carrier settings) to identify which operator is serving, which band, and whether 5G is NSA or SA. This gives you a baseline that is independent of your IoT hardware and SIM. If the phone shows 5G NSA on Band 78 with strong signal, you know the RF environment is adequate. If it shows LTE Band 20 with marginal signal, you know you have a coverage challenge before you start looking at routers or SIMs.

Step 2: Test with a native operator SIM

Before assuming the router or antenna is the problem, test with a native consumer SIM on the operator showing the best coverage at that location. If the router connects to 5G SA on a native SIM but falls back to LTE on the IoT SIM, the problem is SIM provisioning, not hardware. Contact the IoT SIM provider and ask specifically about 5G SA roaming agreements on the operator in question.

Step 3: Verify the antenna configuration

Check that the antenna covers the bands being used by the serving operator at that location. Confirm MIMO port connections are correctly made and that cables are within length limits for the frequencies in use. If using separate MIMO antennas, verify they have adequate physical separation. Check connector integrity – a loose SMA connector introduces return loss that can look like a poor signal environment.

Step 4: Check band locking

On Teltonika and most other enterprise routers, you can lock the device to specific bands or to specific network types (5G SA, 5G NSA, LTE). Testing with different band combinations helps isolate whether the issue is specific to one frequency band (suggesting an antenna or interference issue) or consistent across bands (suggesting a SIM or network architecture issue).

Step 5: Review the device specification against the application

Confirm that the device’s LTE category or 5G capability actually matches what you need. A Cat-4 router connected at Cat-4 speeds is not malfunctioning – it is performing exactly to specification. If Cat-4 throughput is insufficient for the application, the solution is a higher-category device, not reconfiguring the existing one.

The Network Transition and What It Means for Decisions You Make Now

The 5G SA buildout in the UK is real and progressing. EE/BT has been extending SA to additional UK locations through 2025. The Vodafone-Three merger includes substantial 5G investment commitments. O2 is progressing its SA programme. The coverage will not be ubiquitous in 2026 – rural and semi-rural sites will rely on LTE and 5G NSA for several more years – but the trajectory is clear.

At the same time, the legacy network retirement programme continues. 2G is largely gone from the UK. 3G is done. The discussion about LTE sunset timescales is live at operator level, even if public commitments are not yet made in the way US operators have made them. The working assumption for anyone deploying IoT infrastructure with a 7-10 year operational life should be that LTE will be progressively deprioritised and eventually retired during that period.

This creates a straightforward framework for hardware decisions:

For deployments with operational lives of 3 years or less, Cat-4 or Cat-6 LTE devices remain entirely rational choices. The network will support them. The hardware is mature, well-priced, and well-understood.

For deployments with operational lives of 5-7 years, specify hardware that includes 5G capability – at minimum 5G NSA with LTE fallback, ideally 5G SA capable. The additional cost over pure LTE hardware is modest at this point and the operational insurance is significant.

For deployments with operational lives of 7-10 years, the case for 5G SA capable hardware – including RedCap as that device category becomes available in your chosen product lines – is strong. You are deploying hardware that will spend most of its operational life in a world where SA is the primary network architecture.

For antenna decisions: specify multiband coverage from day one. The incremental cost of an antenna that covers 700 MHz through 3.8 GHz properly is small compared to the cost of a site visit to replace a 4G-only antenna when a 5G SA router is installed in two years. Get the antenna right once.

What You Should Take From This

5G is not one thing. It is NSA and SA, with fundamentally different architectures and different implications for IoT SIM provisioning, roaming agreements, and device capability. The 5G icon on a phone tells you almost nothing about whether a 5G IoT device will connect at 5G speeds at the same location.

LTE categories are not interchangeable labels. Cat-4, Cat-6, Cat-12, and 5G RedCap represent genuinely different hardware capabilities with different throughput ceilings, MIMO requirements, and – critically – different network futures. Specifying the right category for the application and the deployment lifetime is a design decision, not an afterthought.

Antennas matter more than most deployments treat them as mattering. Frequency coverage, MIMO isolation, cable loss, and physical placement are all variables that determine the real-world performance of a cellular deployment independently of the hardware or SIM choice. Getting the antenna specification right from the outset is the single most cost-effective investment in deployment performance available.

The network is transitioning. 5G SA coverage is expanding. LTE is on a long but finite retirement trajectory. Decisions made now about hardware lifetimes, device categories, and antenna infrastructure will determine whether IoT deployments remain supported and performant through their intended operational lives – or require costly mid-life hardware replacement because the connectivity assumptions they were designed around no longer hold.

None of this is complicated once you understand the vocabulary. The industry has done a poor job of communicating it clearly. Hopefully this helps.