All 4 UK network operators launched their early 5G networks in some locations during 2019. To look at how “5G” these networks are, we need to first determine what 5G is.
How can we define 5G?
The International Telecommunication Union's (ITU) IMT-2020 minimum requirements form formal, international baseline requirements for a “5G” service. Standards development bodies (such as 3GPP) then develop 5G technologies to align to these baseline requirements.
The minimum requirements from IMT-2020 are envisaged alongside 3 main use-case scenarios; enhanced mobile broadband, ultra-reliable low-latency communications, and massive machine-to-machine communications.
Some of the headline requirements for 5G networks include a minimum peak download speed of 20 Gigabits per second, and a minimum peak upload speed of 10 Gigabits per second, under ideal, error-free conditions, for a single user, operating on a single band. The headline latency requirement is for 4 ms latency in enhanced mobile broadband scenarios, and 1 ms latency for ultra-reliable low-latency communications scenarios. The minimum total channel bandwidth for a 5G service is 100 MHz, and can be provided either as a single carrier, or an aggregation of multiple carriers.
We can generally break 5G networks down into 3 types:
- 4G networks, with 4G cores, 4G radios, and 5G sales and marketing buzzwords. (For example, 5G E in the USA)
- 5G non-standalone (NSA) networks, most likely running 3GPP Release 15 radios, on a 4G/LTE core.
- 5G standalone (SA) networks, most likely running 3GPP Release 16 on radios and core.
Putting aside group 1, where 5G is merely a marketing buzzword, we now need to understand what standalone and non-standalone networks are. Both 5G non-standalone and standalone networks are a form of 5G New Radio (5GNR). 5G New Radio is generally split into two groups - sub-6 GHz frequency networks, and mmWave (> 6 GHz) networks. The UK has only seen commercial deployment of sub-6 GHz networks to date. The USA has mainly seen deployment of mmWave networks, however.
Non-Standalone Networks
All of the UK's 5G deployments to date have been non-standalone (NSA) 5G networks, using 3GPP Release 15 radios, and spectrum in the 3.4 to 3.6 GHz range. They are focusing solely on enhanced mobile broadband (i.e. faster internet speeds for end users). Early consumer 5G handsets currently focus their support on NSA networks.
5G in non-standalone mode is a relatively straightforward evolution from operators’ existing 4G networks - in essence, new 5G radios are connected to existing 4G base stations, to “piggy-back” on their connection back to the network core. The new 5G base stations are used in a mode called 5G NR E-UTRA-NR Dual Connectivity, which is usually just called EN-DC.
Operators are launching NSA networks since this allows them to be early to market with 5G products, and put some of their existing spectrum holdings to use. UK operators participated in the 3.4 to 3.6 GHz spectrum auction in 2018, leading to a demand to put this spectrum to use as soon as possible to increase capacity and speeds.
NSA networks are built over the top of existing 4G networks, and indeed are fully reliant on them for operation. Indeed, EN-DC NSA networks rely on the existing 4G radios for uplink data (from handset to base station), and only use the 5G NR link for downlink. This is why users have generally not seen significant improvements in upload speeds when testing 5G, even on networks using mmWave, such as Verizon USA.
From a technical standpoint, the 5G spectrum is accessed by way of a slight evolution of the traditional carrier aggregation seen in 4G, where multiple LTE carriers in different bands can be combined to provide a faster service. Under EN-DC, a 5G carrier can be combined with this, for supplementary downlink capacity. In 4G, there has been very little deployment of uplink carrier aggregation outside of Asia, and handset compatibility remains limited.
Interestingly, due to the propagation characteristics of the 3.5 GHz band, there are some significant range benefits to using 4G/LTE for uplink signalling. Lower frequency signals generally travel further than higher frequencies, and 4G networks run at lower frequencies than 5G. Some experiments have shown that very significant (~10x) coverage improvements can be achieved by using 4G for uplink, rather than 5G. This will, of course, come with lower headline speeds, but given the power-constrained nature of handsets, and the limited antenna size in a handset, we will likely see lower frequency uplink for some time to come.
Standalone Networks
Standalone 5G networks have not yet reached significant commercial deployment, mostly due to them being reliant on 3GPP Release 16 technology for the 5G core. 3GPP Release 16 is due to be frozen in March 2020, with full completion in June 2020.
In a SA network, a 5G core and radio access network are used. These give more “behind-the-scenes” benefits to network operators, by allowing them to build out new, simpler, radio access networks, and work on more software-based core platforms. These will come with associated cost savings, as well as some new business opportunities to sell slicing and other “enterprise” styled services to businesses.
In a SA network, the 5G radio talks to the 5G core directly. There's no inherent dependency on the 4G network any more, and uplink data goes across 5G. Lower frequencies can still be used for uplink in SA mode, to avoid the poorer propagation of higher frequency signals, coupled with the reduced transmission power of handsets.
To deliver the low latencies promised in ultra-reliable low-latency use-cases, a standalone network will be required. The vision of those designing the standards is that many operators will use NSA as a stepping stone towards building out a SA network.
So, are we 5G yet?
In the UK at least (much of this will hold for Europe, using sub-6 GHz deployments of NSA), we're “partly” 5G. Uplink is still 4G. Downlink can be 5G, in areas with coverage. Coverage is the “gotcha” however, since operators have only rolled out 5G in a (large) handful of larger towns and cities. On the whole though, we're seeing a fair bit of “relatively real” 5G. At least from a standards perspective, of seeing 5G NR used with Release 15 equipment.
The reality, however, if we re-visit the IMT-2020 standards, is that we're still quite a long way from the 20 Gb/s downlink, and 10 Gb/s uplinks required to be “true” 5G. Absent standalone-mode networks, the required latencies are still some way off, as these will require re-architecting of the radio access networks to achieve the promised reductions.
It's important to manage expectations, however - a quick calculation of 1 ms tells us that, travelling at the speed of light, we can reach a maximum distance of 300 km. This assumes unrealistic and unattainable efficiencies (i.e. no queueing delays, no random access time, and no latency introduced by relaying messages between transmission media). If we need a round-trip to complete within the heralded millisecond, that reduces the maximum distance to 150 km. And again, that's based on impossible “complete efficiency”. Given the limited range of radio, we could revise this to the speed of light in glass, and see a round-trip within 1ms would limit us to 100 km range. As the unrealistic assumptions are removed, this will further decrease. These limits are defined by the laws of physics, however - specifically, the speed radio waves and light can travel through air and glass respectively.
Nokia has reported 1 to 2 millisecond latencies from phone to base station. There's still some way to go until we reach the hyped 1 millisecond latency between any two meaningful points.
Conclusions
UK networks are not in as bad a place as some of the US carriers were with their “5G E” networks, which offered no real 5G, and were a rebranding of existing 4G networks.
In terms of “how real” our 5G is, it's as real as it can realistically be right now, but that's a long way off the technical IMT-2020 definition of 5G. It's probably a bit early to be honestly calling the service 5G.
It's worth bearing in mind that this was also true of 4G however - LTE (Release 8) was designed to meet the IMT-Advanced requirements from the ITU, and advertised as 4G, however it was not until LTE Advanced (Release 10) that the IMT-Advanced requirements were actually satisfied, through the ability to reach theoretical gigabit downlink speeds. And even today, we've yet to see gigabit speeds in the UK over LTE, even on an empty cell, with 2x2 MIMO, and carrier aggregation. There's plenty of reasons for this, and those will perhaps be the topic of a future analysis.
