4G SIM NewsIndustrial IoT connectivity has a reliability problem that nobody in the industry likes to talk about plainly. Cellular networks are almost always available. They are not always on. There is a difference, and the M2M and IoT sector has spent twenty-five years building workarounds for that gap.
This post traces the history of IoT SIM cards through the evolution of UK mobile networks – from the analogue duopoly of the mid-1980s through the messy, expensive, genuinely transformative arrival of 3G – and explains how the peculiarities of that history shaped the hardware and SIM strategies that industrial IoT engineers use today.
The UK Mobile Landscape Before M2M Was a Thing
The UK mobile industry started as a duopoly. Vodafone made the first UK mobile phone call on 1 January 1985. BT’s Cellnet network launched six days later on 7 January 1985. For eight years, these were the only two players. Coverage was urban, calls were expensive, and the idea of a machine using a mobile network to send data was not yet a practical proposition.
The first serious disruption came in September 1993 when Mercury One2One launched – the world’s first 1800 MHz GSM network and the UK’s third mobile operator. One2One competed aggressively on price, targeting consumers rather than corporate accounts, and forced Cellnet and Vodafone to respond. A fourth network, Orange, launched in 1994. By the mid-1990s, the UK had what would remain the standard four-network structure for the next three decades: Cellnet (which became BT Cellnet in 1999, then O2 in 2002), Vodafone, One2One (which became T-Mobile in 2002), and Orange.
The transition to GSM 2G across these four networks through the mid-1990s created the first practical data capability on UK mobile networks. GPRS – General Packet Radio Service, the packet data overlay on GSM – arrived in 2000, with BT Cellnet launching the world’s first commercial GPRS network in June 2000. GPRS speeds were modest by any standard (theoretical maximum around 114 kbps, real-world typically 20-40 kbps), but it was enough for early M2M applications: meter reading, alarm signalling, basic telemetry, ATM communications.
This was the era when M2M as a commercial practice began in the UK. A handful of specialist SIM providers started offering industrial SIM cards – often on single-network arrangements with one of the big four – configured for always-on data rather than voice-centric consumer use. The APNs were different, the pricing was different, and the support requirements were different. The term IoT SIM did not yet exist. The term M2M SIM was just emerging. Most people called them data SIMs and had long conversations with network account managers about why their modem kept losing its data session.
The 3G Auction and What It Did to the Market
In April 2000, the UK government auctioned five 3G spectrum licences in what became one of the most financially consequential events in British telecoms history. The five licences raised £22.5 billion. Four went to the existing operators – Vodafone, BT Cellnet, Orange, and One2One. The fifth went to Hutchison 3G, a new entrant backed by Hong Kong conglomerate Hutchison Whampoa.
That fifth licence created Three. Hutchison launched the UK’s first commercial 3G network on 3 March 2003 – a date deliberately chosen to match the brand (03/03/03). Three had no 2G network. It was built entirely on 3G infrastructure, which made it simultaneously the most forward-looking and most coverage-limited network in the country. In those early years, Three maintained national roaming agreements with O2 and later Orange to fill the gaps where its own 3G signal did not reach.
The other four networks launched their own 3G services between 2003 and 2004, though most concentrated initially on business customers. Vodafone launched its consumer 3G service in late 2004. T-Mobile followed in autumn 2003. Orange and O2 completed their 3G rollouts through 2004 and 2005. Coverage outside major urban areas remained patchy for years.
For M2M, 3G brought two things: higher data speeds and, more practically, a more stable data session model than GPRS. UMTS connections were more reliably maintained than the intermittent GPRS sessions of the 2G era. But they were far from perfect. Coverage variation between networks was significant and geography-dependent. A device on Orange might have full signal where the same device on Vodafone had none, and vice versa. For industrial deployments at fixed sites – substations, pump stations, remote cabinets – you discovered your coverage situation when you installed the equipment, not when you planned the project.
Why Network Coverage Was – and Remains – Unpredictable
Understanding why M2M connectivity has always been unreliable requires understanding how cellular coverage actually works in practice, as opposed to how it looks on network coverage maps.
Coverage maps show theoretical signal at ground level in open conditions. Real deployments happen in metal cabinets, plant rooms, buildings with reinforced concrete floors, locations surrounded by steel racking, or sites in terrain shadows. The signal that reaches the SIM inside the equipment is not the signal that reaches a phone held at head height in the same postcode.
Beyond physical attenuation, there is network congestion. A mast that handles residential and commercial traffic runs at very different load levels at 3am versus 8am. An M2M device trying to send telemetry data at 8am may find the mast it connects to contended. It completes its data session eventually, but with latency and potential session drops that a consumer using a phone app would never notice but a SCADA system logging to a fixed interval would.
Then there is network maintenance, firmware updates, and outages. Every operator has had significant outages. O2 had a 24-hour data outage in December 2018 affecting up to 32 million users. EE, Vodafone, and Three have all had major incidents. No network has a perfect uptime record, and M2M devices on single-network SIMs inherit their operator’s downtime as their own.
The M2M sector’s response to all of this was not to wait for networks to become more reliable. It was to build redundancy into the solution architecture.
The Rise of Multi-Network SIM Cards
The first generation of multi-network IoT SIM cards appeared in the mid-2000s. The concept was straightforward: rather than locking a SIM to a single operator, the SIM could roam across multiple UK networks, connecting to whichever provided the strongest signal at the deployment location.
In practice, early multi-network SIMs were often built on roaming agreements rather than native multi-network access. The SIM had a home network, and when that network was unavailable, it would roam to an available alternative – subject to the roaming agreements in place. This worked, but with latency in the handover and occasional situations where the roaming logic did not behave as intended.
Proper multi-network SIM cards – sometimes called multi-IMSI or steered roaming SIMs – carried multiple operator identities on the same SIM and could switch between them based on signal quality and network availability, without going through a traditional roaming process. These became the foundation of serious industrial IoT SIM infrastructure through the 2010s.
The value proposition was clear: one SIM, coverage across EE (which absorbed T-Mobile and Orange in 2010 and launched the UK’s first 4G network in October 2012), O2, Vodafone, and Three. Where one network had a coverage gap or an outage, the SIM connected to another. For deployments at fixed sites where you could not predict which network had the best signal, multi-network coverage was a significant practical advantage over single-network SIMs.
4G and the Consolidation Era
EE launched the UK’s first 4G LTE service in October 2012, formed from the merger of T-Mobile and Orange. It was a significant moment – the two brands that had competed for UK consumers since 1993 and 1994 disappeared into a single entity that became, under BT’s ownership from 2016, the UK’s largest mobile network by coverage.
O2, Vodafone, and Three followed with their own 4G services through 2013. LTE brought substantially improved data rates and, critically for M2M, more reliable data sessions and lower latency than 3G. The M2M sector migrated from 3G to 4G through the mid-2010s, and 4G became the standard technology for industrial IoT connectivity for the rest of the decade.
Network sharing arrangements changed the infrastructure map. O2 and Vodafone formed the Cornerstone joint venture, sharing tower infrastructure. EE and Three formed the MBNL joint venture for similar reasons. Effectively, four networks began operating from two physical tower infrastructures, which created some convergence in their coverage footprints – useful context for understanding why coverage testing across all four networks still produces meaningfully different results.
The 3G switch-off is now underway. Vodafone and EE completed their 3G switch-offs in early 2024. Three completed its 3G shutdown in late 2025. O2 is expected to complete its switch-off in early 2026. For the M2M sector, this creates a migration challenge: devices deployed on 3G-only hardware need replacing, and any SIM configured with 3G as its primary network access technology needs reconfiguring. The 2G sunset – confirmed for 2033 at the latest – creates a longer runway, but the same eventual requirement.
Why Dual SIM Routers Were Invented
Multi-network SIM cards solved the coverage problem at the SIM level. But they did not solve the network failure problem at the hardware level.
Consider the scenario: a single-SIM industrial router at a remote monitoring site. The SIM is on a multi-network roaming plan, so it has good coverage. But the SIM’s profile management system has a temporary issue. Or the private APN has a billing lapse. Or the SIM itself develops a fault – a micro-crack in the contact, a corrupted profile, an antenna connector that has worked loose over two winters of thermal cycling. The device goes offline. The multi-network SIM is no help if the SIM itself is the failure point.
The dual SIM router addressed this by providing two completely independent cellular connectivity paths. Two SIM slots, two independent data connections, two different operators – either or both could be active, and the router would switch automatically between them based on configurable failover criteria: signal strength falling below a threshold, the primary connection failing a ping test, data limits being reached, or simply on a scheduled basis.
Teltonika’s RUT950 and RUT951 built much of their reputation on this capability in the UK market. The current generation – the RUT901 with four Ethernet ports and dual SIM, and the RUT956 with dual SIM, GPS, and RS485/RS232 serial ports – carries forward the same logic. For the full picture on how these models fit together and which application suits which router, the dual SIM router guide at 5grouters.co.uk is a useful reference.
The practical failover scenarios that dual SIM handles include carrier outages, mast maintenance, SIM profile issues, data allowance exhaustion on one SIM, and geographic coverage variation for mobile deployments. The two SIMs do not need to be on different networks – though putting them on different operators is the highest-resilience approach. Some deployments run a fixed IP SIM for primary access and a roaming SIM as backup. Others run two different multi-network SIMs from different providers.
Always Available, Not Always On: The Core Problem
The phrase that best describes M2M cellular connectivity is borrowed from how network engineers think about reliability: always available, not always on.
The network is available in the sense that signal is present and the connection can be established. But the active data session – the state in which a device is actively connected and exchanging data – is fragile in ways that wired connections are not. Sessions drop. DHCP leases expire. Routing table entries time out. The carrier’s core network decides a device that has been idle too long has disconnected and releases its resources.
From the device’s perspective, the cellular module may believe it is connected when the carrier’s core network has already torn down the session. The module shows registered and attached. Signal bars look fine. But data does not flow. This is sometimes called a silent failure, and it is genuinely common on long-running M2M deployments.
The conventional response in the M2M sector is the ping test. The router is configured to periodically send an ICMP ping to a known-reliable external address – typically a public DNS server. If the ping fails, the router knows the connection is not functional despite appearing connected, and it takes corrective action: reconnecting the cellular data session, switching to the backup SIM, or rebooting the cellular module.
The more aggressive version of this is the automatic reboot. If ping-based reconnection attempts fail to restore connectivity within a defined timeout, the router reboots itself. This sounds crude, and in some sense it is – but it works. A router that reboots itself after 30 minutes of failed connectivity is far more useful in an unattended remote deployment than a router that sits in a failed-but-connected state indefinitely, waiting for a maintenance visit.
Dedicated tools and monitoring services for exactly this scenario have emerged from the M2M sector. Ping Reboot is one example – a purpose-built connectivity watchdog that monitors device availability and triggers automated recovery actions when connectivity fails. For deployments where the router is inside a cabinet that cannot easily be power-cycled, where the cellular module has a known tendency to hang, or where SLA requirements make any unplanned downtime significant, automated watchdog tools are part of the standard deployment toolkit.
The IoT SIM Card Matures: Fixed IP, Private APN, and Pooled Data
Through the 4G era, IoT SIM cards evolved significantly beyond the basic multi-network roaming capability of the early M2M SIM.
Fixed IP SIM cards – providing a static, publicly routable IP address for the device – became a standard requirement for any application where inbound connections needed to be made: remote access to SCADA systems, camera viewing, VPN endpoints, any scenario where a central system needed to reach out to a remote device rather than waiting for the device to initiate contact. Fixed IP SIMs on private APN infrastructure provided this without the security exposure of a public IP on a shared carrier network.
Private APNs – dedicated access point configurations that route a customer’s device traffic through private infrastructure rather than the public internet – addressed the security requirements of critical national infrastructure, financial applications, and enterprise deployments where data sovereignty and network isolation were non-negotiable.
Pooled data plans changed the economics of fleet deployments. Rather than each device paying for a fixed monthly allocation regardless of actual usage, a pool allows a fleet’s total data consumption to be managed collectively. A device that sends a burst of telemetry data when something goes wrong draws from the pool without incurring overage charges on its individual allocation, while devices that are quiet in a given month leave allowance available for devices that need it. For fleets of hundreds or thousands of devices, the cost difference between pooled and per-device data plans is material.
The Present: Multi-Network Roaming, eSIM, and What Comes Next
The UK IoT SIM market in 2025 is more sophisticated and more competitive than it has ever been. Multi-network roaming across all four UK operators is a commodity feature. Fixed IP and private APN options are widely available. eSIM with remote profile provisioning – eliminating the need to physically access a device to change network provider – is becoming standard in new hardware.
Vodafone and Three completed their merger in 2025, creating VodafoneThree – a combined entity with the UK’s largest spectrum holdings and the most extensive 5G coverage. This reduces the UK from four independent mobile networks to three: VodafoneThree, EE (BT), and Virgin Media O2. The long-term implications for multi-network SIM roaming arrangements and pricing are still working through the market.
5G is expanding the performance envelope for M2M applications that were previously constrained by 4G bandwidth: high-definition CCTV with cloud recording, real-time video analytics, dense sensor networks in smart cities and industrial environments. But 5G does not eliminate the fundamental reliability challenge. The always-available-not-always-on problem persists regardless of air interface generation.
What has changed is the tooling available to manage it. eSIM with SGP.32 remote provisioning, combined with platforms like Teltonika RMS for device-level monitoring and automated recovery, creates a management infrastructure that the M2M engineers of the GPRS era could not have imagined. The hardware, the SIM technology, and the management platforms have all matured. The underlying cellular network remains, as it always was, the least predictable component in the chain.
For anyone specifying IoT SIM cards for current deployments – whether that is a multi-network roaming SIM for a single-SIM router like the Teltonika RUT200, a dual-SIM configuration for a RUT901 or RUT956, or an eSIM solution for sealed long-life installations – the fundamentals have not changed: match the SIM capability to the uptime requirement, test the actual coverage at the deployment site rather than trusting a map, and build automated recovery logic into the router configuration from day one.
The networks are more reliable than they were in 2003. They are still not reliable enough to deploy without a plan for when they fail.