South Korea was an early leader in 5G roll-out but has lagged behind recently over deployment of millimeter wave spectrum in the 28 GHz band. It seems almost as though dented national pride lay behind the cancellation of two mmWave spectrum licences by the country’s government, those of Korea Telecom (KT) and LG Uplus, respectively the number two and three MNOs in the country with 55% of subscribers between them.
Even SK Telecom, the biggest with almost all the remaining 45% of subscribers, was censored and had its 28 GHz licence term reduced by six months. SK was also warned it would lose its licence altogether if it failed to meet its original commitment of establishing 15,000 base stations using the spectrum by the end of May 2023, the same target as the other two were set.
All three operators acquired 28 GHz spectrum in 2018 at a time when South Korea was forging ahead in a blaze of 5G optimism, claiming to launch the world’s first commercial 5G network in April 2019, just a whisker ahead of Verizon in the USA. The mmWave bands were then seen as the enabler of ‘true 5G’, delivering the maximum promised performance and capacity, and under that aura the three Korean operators made deployment promises more extravagant than their counterparts almost anywhere else, in return for the spectrum awards.
All three operators also acquired midband frequencies in the 3.5 GHz band at the same time and roll-out in that spectrum has proceeded apace, with all three meeting their commitments on that front. But it has turned out that the midband 5G network has comfortably met demand for capacity and performance so far, leading the operators to hold back on mmWave roll-out. Whether they had calculated on losing the spectrum rights altogether is unclear, nor has SKT revealed whether it intends to ensure it meets its 2023 commitment.
As it is, the government decision is awaiting finalization in December 2022 at a hearing, after which the operators would be forced to shut down their network infrastructure that uses the mmWave spectrum. The government has hinted it will then seek to bring in new players as part of a plan to introduce more competition into the country’s mobile field.
South Korea’s mmWave setback may only be temporary and does not reflect a long-term shift away from higher frequencies, which after all are likely to underpin the future evolution to 6G. It does reflect a realization that mmWave is still work in progress, with research ongoing in technologies for mitigating the signal degradation that occurs with uncontrolled transmission, while there has also been a change in thinking over how best to license the spectrum. South Korea is to some extent a victim of its early foray into mmWave at a time when it was assumed that, with adequate densification, higher speeds could be attained across large areas with relatively little variation within cells.
Verizon, an early adopter (though initially for fixed wireless) also championed mmWave but then reined back, as it turned out that its subscribers could only obtain the full benefits within relatively small areas. The mmWave propagation fading was more serious in urban environments than had been anticipated because of building penetration loss, as non-line-of-sight (NLOS) transmission increases diffraction and reflection significantly. For this reason, these losses also impair indoor penetration.
After some earlier trials maintained hopes that Korean mmWave roll-out would proceed at least in line with the government’s minimum stipulated target, it was becoming increasingly clear early in 2022 that LG Uplus and KT were well behind and expressing increasingly negative sentiments about the band. By then, 45% of the country’s population had access to 5G, but very few to mmWave, since only about 4,000 base stations, just 2% of the 215,000 total for 5G, supported it.
KT’s Ku Hyun-mo complained to Reuters in May 2022: “Rolling out 5G that is 20 times faster is nearly impossible, even in Seoul. Establishing nationwide coverage just can’t be done. The fastest version millimeter wave spectrum travels straight and it can’t go around obstacles. It can’t deliver the same speed once it travels a few hundred meters.”
Yet the Korean government is now pointing to Japan and the USA as examples of where South Korea should be with mmWave. This is somewhat misleading, because all is never as it seems with mmWave and every country’s situation is different.
In the USA, mmWave has been tilted towards fixed wireless access (FWA) in the consumer arena, and towards industrial applications for enterprises. This is especially true since AT&T and Verizon secured C-band spectrum, giving them the midband assets they had lacked at the start of 5G – a shortfall that forced them to turn to the more challenging mmWave frequencies to add capacity.
More generally FWA has become the main vector of growth for mmWave in the immediate term, as has been acknowledged by the GSA, which forecasts that FWA CPE shipments will grow by 33% in 2022 to 21.9m units, after subsiding in 2021 because of the Covid-19 pandemic. A substantial proportion of this CPE supports mmWave.
Japan is one country where mmWave roll-out has stayed more or less on track, with the operators sustaining their enthusiasm, but even there, commercial deployments have lagged earlier expectations. The emphasis there has shifted somewhat from performance to capacity in the country’s many dense urban environments. In this guise mmWave takes up the slack in dense urban cells by providing extra capacity without necessarily offering a guarantee of higher bit-rates.
This requires dual connectivity as supported by 5G NR, in which two base stations, which may or may not be colocated, enable carrier aggregation. The midband base station operates as a master node in the FR1 (midband) spectrum and the mmWave base station at 28 GHz acts as a secondary node in FR2.
Between them the Japanese operators – NTT Docomo, KDDI, Softbank and Rakuten – have deployed just over 20,000 mmWave 5G base stations, which is less than half the very ambitious 45,000 target set in South Korea (which has less than half the population of Japan) and which the operators have failed to meet. The Japanese operators have focused on hotzones in Tokyo and other cities, targeting the most congested areas to alleviate the burden on the midband.
Services were launched this year and Japanese consumers can now purchase mmWave smartphones from Samsung, Sony, Sharp, Fujitsu and Google, with other suppliers coming online with lower-priced models. Qualcomm, as supplier of silicon for the phones, has been in the vanguard of this roll-out, conducting tests to compare performance between mmWave and midband, and talking up the former.
Qualcomm has cited 2Gbps download speeds, and above 300Mbps upload speeds, on average around six times and five times faster, respectively, than a premium non-mmWave smartphone would obtain on a given location on the same network. But again, not all locations can access mmWave at those speeds at this stage.
Beyond these countries mmWave is progressing at a more nuanced pace with a realization that the capabilities require more work on deployment and spectrum allocation, while waiting for use cases demanding the extra performance, such as mobile gaming and home robotics, to mature.
This is very much the case in the European Union, where the average country has assigned just 290 MHz of mmWave and many member states have not assigned any, according to the European Commission’s 5G Observatory.
The UK is also at an early stage, with regulator Ofcom publishing proposals only in July 2022, to offer a combination of citywide and local licences. For major towns and cities where the first mmWave deployment is expected in high density areas, as in Japan, Ofcom plans to assign local licences on a first-come, first-served basis in the lowest 850 MHz of the band, using its shared access licensing framework. Then the upper 2.4 GHz of the band would be auctioned to local operators across the whole metropolitan area.
Ofcom has determined this local partitioning will best reflect the varying needs and RF conditions of different areas. Then for sparser areas Ofcom is proposing to assign local licences on a first-come, first-served basis for all of the 26 GHz band through the shared access licensing framework.
Notably, Ofcom proposes to revoke existing fixed-link licences for the 26 GHz band in and around high-density areas, with five years’ notice, aiming to avoid interference with emerging mmWave services in that band. Fixed links that operate in low density areas would however be allowed to remain in the band on the basis that interference issues could probably be mitigated in those cases.
As for the 40 GHz (40.5-43.5 GHz) band, which has been less widely deployed so far, Ofcom has already assigned the bandwidth in blocks to three operators – Hutchinson Three UK, MBNL (the infrastructure joint venture between Three and BT), and MLL 40 GHz (MLL). These assignments are for wireless backhaul or point-to-point links, and not yet for mobile use. Ofcom is therefore consulting on options to make spectrum in this band available for 5G.
Germany, meanwhile, has been developing a new model for licensing mmWave spectrum, having determined from the outset to proceed cautiously in almost the opposite of the South Korean approach. Regulator BNetzA has proposed an alternative to auctions for mmWave spectrum, having made 3.25 GHz of spectrum (24.25–27.5 GHz) available for a fee to any entity with a clear plan for its utilization. The fee is determined by the amount of spectrum requested, the population and area covered, and the duration, with the only condition then being that the spectrum must be used within a year.
In keeping with the German approach to 5G generally, enterprises and small ISPs are invited to apply, as well as large operators, in this case with the option of taking nationwide or local licences. The thinking, perhaps, is that an established MNO might license spectrum for a city center, an ISP for high-speed Internet in lower-density suburbs, and enterprises within their campuses.
In parallel with these moves in licensing, developments of techniques to improve mmWave signal propagation in NLOS conditions continue, aiming to reduce the performance fade-out with distance from the base station that has put off Korea’s operators. Beamforming is one technique, which has fairly obvious gain when line of sight is available, because the energy can be focused in the direction of the receiver.
However, beamforming can also improve NLOS propagation by turning reflection and refraction into an advantage when obstacles block direct signal propagation. This can also improve mmWave coverage greatly during normal roaming within a cell, when a device’s RF environment changes constantly as a previous propagation path becomes blocked when some obstacle comes into line. The idea then is to exploit multiple beams from as many directions as possible through signal processing techniques that overcome offsets in frequency or phase.
The high mmWave frequencies have one advantage for propagation, the shorter wavelength, which means that smaller antenna can transmit and receive signals. This allows incorporation of multi-antenna arrays even in CPE, which in turn brings scope for adaptive real-time beamsteering.
Simulations and field testing have shown that, even with standard omnidirectional antennas radiating equally in all directions, it is possible to capture reflected signals more effectively at the higher mmWave frequencies than in the midband, which means that NLOS signals can supplement LOS signals to increase channel capacity. Indeed, some tests have shown that while the NLOS signal tends to be weaker than a LOS signal if available at midband frequencies, it can be stronger in the mmWave, presumably because reflection characteristics are more favorable.
This has been shown in 28 GHz, 38 GHz, 60 GHz and 73 GHz bands, where objects including building walls, lampposts, tree and foliage can be very reflective, allowing multipath signals to propagate. There is also potential to take the obvious step towards relying purely on reflected signals to maintain an effective link between base station and CPE, without LOS at all.
Several other techniques are evolving to enhance the reflective effect, including Reconfigurable Intelligent Surfaces (RIS), which is expected to be significant in the evolution towards 6G (which is even anticipated to use sub-THz spectrum above 100 GHz). RIS can reflect the radio signal in a programmed direction, independently of antennas.
A RIS can be likened to a repeater, except that it redirects a signal rather than amplifies it, with scope for reorientation in the manner of a mirror to change the direction of reflected light rays. A RIS can be integrated in an advertising panel or even a building’s window glass, being optically passive and so not requiring any power.
These and other emerging techniques will help reduce the rapid drop-off in received power, and therefore data rate, that comes with distance at higher frequencies, and it is clear that in time mmWave will deliver on its promise. It is just that, as South Korea’s operators discovered, hype ran ahead of experience in the early days of 5G.