One overused phrase that has thankfully largely fallen out of usage in the 5G era is ‘spectrum famine’. Despite the high bandwidth and resulting speeds envisaged for 5G applications, there has been concerted effort by regulators and technologists to maximize the spectrum that the new radios can use, and to start to lower the barriers to using it.
Cristiano Amon, president of Qualcomm, is one executive who has predicted that, while about 1,000 band combinations are possible in 4G (which has 49 official spectrum bands) that figure will be about 10,000 in 5G, making the current model of global band harmonization redundant, and putting dynamic spectrum usage and software-defined radio center stage.
The challenge for 5G is not lack of spectrum, but how to knit together multiple assets in the most seamless way; and to combine low with high bands, and exclusive with shared spectrum, to achieve the highest quality of service for the lowest outlay. In other words, spectral efficiency is top of the agenda, to ensure that plentiful spectrum does not equate to wasted or underused spectrum.
The key developments that have ended the famine have been:
- Emergence of affordable, CMOS radios for millimeter wave bands
- Improved techniques, such as spectrum access systems (SAS) to enable 4G and 5G services to share with incumbents in bands such as the USA’s CBRS or the UK’s planned moves for new entrants to be able to harness underused channels in MNO bands
- Development of 4G, and in future 5G, standards allowing cellular networks to run in unlicensed or shared spectrum (LTE-LAA, MulteFire, future 5G-Unlicensed)
- Increased regulatory tendency to issue geographically localized licences, at low or no cost, to encourage specialized build-outs for industries or cities
- Beginning of the emergence of dynamic spectrum usage, so that different technologies can use the same spectrum interchangeably; and so that services or users can access the same bands on an as-needed basis – eventually, potentially, on a pay-as-you-go basis.
The risk is that the impact of all these developments will be lessened because operators will end up with a mish-mash of spectrum which they will utilize in a haphazard and wasteful way. They need to be able to use the best, and most cost-effective, frequency for a particular application at any given time, and to combine low, midband and high frequency spectrum to deliver the best user experience for the best price. This requires some significant further technical work – primarily in automation of dynamic spectrum management, to achieve the vision of ‘always best connected’ at last; and in devices and RF chipsets that can manage large numbers of spectrum bands in a flexible, power-efficient and seamless manner.
All the new sources of spectrum have some downsides (for MNOs, anyway) compared to exclusively licensed frequencies below 3 GHz, with their easily manageable QoS and good balance of capacity and coverage. The key to 5G success will be to address the individual challenges of each one, but more importantly, to evolve a dynamic, intelligent and automated system that will orchestrate all the spectrum available, and form combinations so that the challenges of one band are offset by the qualities of another.
Some of the techniques which are already in use, or are emerging in the near term to improve spectrum efficiency, include:
- Using millimeter wave spectrum – the primary source of the new spectrum plenty – for high capacity hotspots, while maintaining a mobile connection over wider areas using sub-GHz bands, and this is already supported in many chipsets and devices. Already, there is discussion, among academics and politicians, about ‘6G’, whose R&D focus is currently on even higher frequency ‘terahertz’ spectrum, to boost capacity and bandwidth still further.
In the short term, however, mmWave certainly does not address all the issues that sometimes held 4G back from delivering its full impact. It may take us a step closer to the kind of bandwidth and assured quality of service that is needed for fully immersive experiences, but its economics are challenging. Its short range and weakness at penetrating walls means it is best used for very dense hotspots of high usage, often indoors, which might support an immersive experience for a relatively stationary user, in a stadium perhaps, but not a device which is on the move over wide areas.
Using lower frequency spectrum for coverage, to keep the video stream connected while the user moves between hotspots, is a good compromise – but it is a compromise, given that the QoS will decline significantly if the device moves into sub-GHz, or even midband, spectrum.
- Dynamic spectrum sharing (DSS) is a long way from the full vision of as-required, automated access to the best available connection, but it is important to allow operators to make the best use of all their spectrum assets by allowing 4G and 5G (and others) to use the same frequencies. It also provides a more nuanced choice of strategy for operators rather than a stark decision of whether to switch off legacy networks completely to refarm the spectrum. DSS, and precursor technologies like Huawei’s cloud-based spectrum management, are good stepping stones.
- The long-established practice of configuring devices and networks so that users are offloaded to WiFi in unlicensed spectrum, reducing load on the cellular network and improving user experience, is now being extended to cellular connections in shared or unlicensed bands. Devices to support this approach are starting to appear for the USA’s CBRS band, and for gadgets that support access both to low power IoT networks and mobile broadband.
These existing techniques will be refined and expanded over the coming years, and building on the current small steps will be critical to enabling 5G (and 4G expansion) to fulfil their maximum potential in terms of the variety of services, the quality of experience and the profitability of the operators.
Not all the barriers to that nirvana are technical or commercial. Regulatory practice must change too. For every regulator that is adopting a more multi-layered and flexible approach to spectrum, such as France’s ARCEP, or the USA’s FCC, there is another that is persisting with the old approach of auctioning relatively small amounts of spectrum for very high prices – a practice that, in India, may even lead to an auction boycott, or the exit of the largest player, Vodafone Idea.
And while spectrum sharing, and the combination of unlicensed and licensed bands, are very important developments in spectrum efficiency, they do not come easily. Incumbent users of bands which are targeted for cellular usually defend their turf assertively, as the world is seeing in the USA’s efforts to free up spectrum – first in the 600 MHz TV bands, where the incentive auction was held after years of wrangling between broadcasters and the FCC; and now in the C-Band around 3.5 GHz, which is occupied by satellite providers. Even in established licence-exempt bands like 60 GHz, there are efforts to protect quality of experience by limiting sharing, as high bandwidth 5G and WiFi 6 applications threaten future congestion even in these high capacity areas of the spectrum.
Regulators, incumbents, operators and chip vendors will need to cooperate to an unprecedented extent to address these issues, accelerate the availability of new spectrum, and ensure that the end of the spectrum famine is a permanent one.