15 July 2019

5G in five (not so) easy pieces

Tom Wheeler

Tom Wheeler served as the 31st chairman of the Federal Communications Commission from 2013 to 2017. This paper is adapted from a presentation made at the request of the Government Accountability Office.

Throughout the world, ink is being spilled and electrons exercised in a frenetic focus on fifth generation wireless technology, or 5G. The 5G discussion, with all its permutations and combinations, has grown to resemble an elementary school soccer game where everyone chases the ball, first in one direction, then another.

In classic network engineering terms, the “noise” surrounding 5G is interfering with the “signal” about just what 5G is and what is necessary for its introduction. Consideration of 5G is far more serious than the so-called 5G “race” concocted by those seeking to advantage themselves in the business or political market—especially the political market.

There are five often misunderstood facts to know about 5G:
5G is revolutionary because it replaces the hardware components of the network with software that “virtualizes” the network by using the common language of Internet Protocol (IP).
5G is evolutionary as both its new radios and the core network functions are defined as a progression from 4G. Like 4G before it, in most markets 5G will roll out in steps.
5G is not transformational, per se. What will be transformative are the applications that will use the network. The United States was not the first to deploy any of the “G’s” of wireless networks, but nonetheless dominates the wireless ecosystem because of the innovative technologies developed by American entrepreneurs for those networks.
5G is a cybersecurity risk because the network is software based. Earlier networks’ reliance on centralized hardware-based functions offered a security-enhancing choke point. Distributed software-based systems, per se, are more vulnerable.
5G is spectrum dependent. In the long run this means new spectrum allocations. While those are underway, however, the evolution has begun using old spectrum assignments.

“Winning 5G” is not so much a “race” as it is a process. Characterizing 5G as a contest demeans its great technological progress and the policy challenges that progress presents. 5G should be more than a political talking point; the new network represents the need for a meaningful policy strategy.

What is missing in today’s 5G policy discussion is a focused identification of deliverables that go deeper and are more meaningful than the ill-defined “winning” of a so-called “5G race.” To consider this, let’s parse 5G into five (not so) easy pieces:
What will it do?
What is the technology and what about spectrum?
What about cybersecurity?
What are the hidden issues?
Does first to the tape really matter?

To listen to its proponents, 5G is on par with the genius of Edison. For equipment manufacturers it is a new revenue stream at a time when the 4G market was becoming saturated.[1] For wireless carriers, it similarly offers a vision of high margin network-based applications in place of today’s commoditized carriage of other companies’ data. And for political actors, 5G is the perfect excitement vehicle: something new and better, the promise of which could be theoretically lost without the actor’s decisive leadership.

Perhaps, however, we should listen to Bill Stone, Verizon’s vice president of technology development and planning. Wisely, Stone recently warned, “There is a potential to overhype and under-deliver on the 5G promise.”[2]

There is no question as to the importance of 5G. It is the most significant network overhaul in history because the alchemy of digital technology allows the transformation of what was always done in hardware to become functions accomplished in software. Then, with such a virtualized network, the power of the lingua franca of Internet Protocol (IP) takes over to eliminate the need for specific technology protocols for specific functions.

Yet, while 5G holds great promises, the wisdom of Stone’s admonition should be our North Star. It is time to take a deep breath and realize that 5G is something more than marketing slogans or technology to be weaponized for political purposes.

“It is time to take a deep breath and realize that 5G is something more than marketing slogans or technology to be weaponized for political purposes.”

As a part of the lead up to the 2016 allocation of the world’s first 5G-dedicated spectrum, I told the National Press Club, “If anyone tells you they know the details of what 5G will deliver, walk the other way.”[3] Insofar as all the potential new services 5G could enable, it is an admonition that still stands three years later. 5G will be both evolutionary and revolutionary. At first, consumers will notice only incremental improvements as 5G is used principally for capacity expansion. Over time, however, an all Internet Protocol (IP) based network will open new possibilities.

5G is a vast framework for the networked application of spectrum. When the functions of the network are virtualized in software, the nature of the network is transformed from its traditional role of transporting information between points, to abstracting and orchestrating digital information within an all-digital network. Because it is a software-driven network, 5G may be the last physical network overhaul in generations as upgrades will now be only a matter of replacing software and low-cost, commodity components.

The details of the new applications that will use the network are still in the imagination stage. How they will function, however, is not. The ubiquitous Internet Protocol will be the language of both the network architecture and the applications that run on it. Thanks to IP, 5G will be able to run multiple concurrent application layers—each tied together by IP—as opposed to legacy telecom networks that could only perform tasks sequentially. This capability ignites innovation potential while exposing 5G to the vast array of resources inherent to cloud computing.

Just what collapsing everything to IP (as opposed to the tradition of unique protocols for unique applications) means in terms of applications is up for grabs. It is revealing, for instance, that Verizon is running a crowdsourced “Build on 5G Challenge” contest in which they solicit ideas for 5G products, services, or applications.[4] Answering this question will be crucial to how and at what speed 5G is deployed. As one trade publication put it, “operators are looking for the ‘killer’ 5G app because the cost of rolling out 5G networks is expensive, and the costs won’t be recouped by simply increasing prices for mobile connections.”[5]

While innovators explore the promise of an all-IP world, early mass market applications of 5G will probably be more pedestrian. One industry analyst suggested that, like other internet experiences, the high-speed, low latency realism of 5G might make pornography an early moneymaking application.[6] Another analyst suggested that real time gambling during sporting events will take advantage of 5G’s high-speed, low latency connections.[7]In other words, if previous internet history holds, when consumers slip on 5G-enabled AR and VR headsets, it may not just be for a tour of Tutankhamun’s tomb.

AT&T’s first 5G customers have been businesses that used it as a replacement for their wired local area network (LAN).[8] Verizon’s first 5G customers are what one might call “wireless cable”—a service that delivers high-speed broadband capacity to homes in competition with the local cable company. The ability to handle a multitude of Internet of Things (IoT) connected devices has also been promoted as a 5G deliverable. While most IoT applications don’t require high bandwidth, even at slow data rates, thousands of IoT devices communicating at the same time will require 5G capabilities.[9]

While the all-IP network enables Buck Rogers-like potential, at this point there is one driving force behind 5G: the application of the new technology to expand network capacity.

Thanks in large part to the ever-growing demands for wirelessly delivered video, U.S. carriers are forecast to run out of available 4G spectrum as we approach 2022.[10] Today, for instance, about three-quarters of mobile network data is used for video traffic.[11] Disney, for instance, reports that over 70% of its content is today accessed via a mobile device (up from 40% four years ago).[12]

The Cisco Visual Networking Index forecasts a greater than threefold jump in global mobile video traffic from 2019 to 2022.[13] Buck Rogers will come, but first 5G will be more of the same. As the industry analysts at New Street Research put it, “the primary role of 5G will be to enable operators to continue to add capacity at acceptable costs.”[14]

And it is great capacity: fast, and with low latency (the time elapsed waiting for a response). The obvious use of 5G is to provide capacity to fill the spectrum demands for what we know now: video. There is nothing to be ashamed of here. A study by 5G manufacturer Ericsson found that 5G adoption would come in three phases. The first phase would be premium smartphone downloads of content (usually this is video content) in seconds rather than minutes. This would be followed by 5G home wireless broadband to challenge traditional cable TV (video and broadband delivered video). The final phase Ericsson predicts, is 5G hot zones of ultra-high speed in airports, offices and shopping areas.[15]

While 5G will drive mobile video services, any consideration of what 5G will be must take notice of the immutable lesson of network history: that it is not the primary network that is transformative, but its secondary effects. “If we’ve learned anything in the generational march through wireless connectivity,” the 2016 Press Club remarks reminded us, “it is that we have always underestimated the innovation that would result from new generations of wireless networks.” 3G networks, for instance, were built on an economic model that did not anticipate how the iPhone would change the nature of those networks. The technology of 5G networks far exceeds the technological upgrade from 2G to 3G—yet, looking at what that earlier upgrade unexpectedly enabled, we cannot underestimate what’s next. The immutable law of network history will again repeat itself as the 5G network spawns transformational secondary effects.

The technical standards for 5G remain a work in progress. The international standards-setting group 3GPP operates by periodic “Releases” that roughly coincide with network generations. Release 15 includes many key functionalities, which some would consider low-hanging fruit. However, Release 15 has been pushed back by three months into the third quarter of 2019; final release will come mid-2020. The first iterations of Release 16, which will contain key standards necessary for IoT, will not be available until 2020 with the final standard some significant time after. Even after a standard is released, it takes years for it to be widely implemented.

5G is the combination of innovative radio and core network technologies. Everything is now reduced to data, and since IP is the single language of 5G, everything simply becomes an IP app. There are three unique but interrelated factors at the heart of the 5G standard: greater spectral efficiency, greater spectrum pathways, and the distribution outward of core network functions.
Greater spectral efficiency

The greater the spectral efficiency—measured in bits per hertz—the greater a cell site’s capacity. A 5G network puts more bits through a given amount of spectrum by: (1) Optimizing the signal-controlling overhead, (2) Thereby making more of the spectrum available for traffic, (3) Mitigating degrading radio interference, and (4) The use of multiple-input/multiple-output (MIMO) antennas that break a transmission into multiple streams across a common channel. By one estimate, using these technologies expands the spectral efficiency of mid-band 5G by 52% over 4G.[16]
Greater spectrum pathways

While spectral efficiency gets more throughput for each hertz of spectrum, the more hertz that can be used, the greater the total throughput. This is accomplished by aggregating existing spectrum assignments and/or by the allocation of new greenfield spectrum. The first is a function of carrier spectrum management, the second a matter of government spectrum management.

Existing mobile spectrum assignments are typically 10 to 20 MHz blocks. The larger the amount of spectrum available, the more data that can be driven through it. Thus, a concept that has been the backbone of wireless technology since the earliest days of cellular has been combined with the basic concepts of a distributed packet network to introduce “carrier aggregation” that virtualizes a broader pathway. One channel of spectrum is used to control the other channel(s) that deliver the content; then the information is broken into pieces to be sent over multiple different pathways and reassembled at the destination. In 4G-LTE—the first evolutionary step toward 5G (LTE standing for Long Term Evolution)—carrier aggregation can bond up to five 20 MHz channels into a virtually expanded pathway. By thus decoupling the control of the network from the carriage of data, 5G makes each independently scalable to further increase throughput.

There is, of course, management overhead in carrier aggregation. The optimal solution is to remove that overhead by making available large swaths of spectrum that can carry large amounts of data. Using greenfield spectrum for 5G, up to five 100 MHz channels can be aggregated for speeds up to 20-times faster. This is what makes the millimeter wave high-band spectrum above 6 GHz attractive: the availability of hundreds of MHz in one place.

Unfortunately, however, all spectrum is not created equal. No one has repealed the laws of physics. The higher you go in frequency, the more oxygen eats at the signal to reduce the distance it can travel, and the more that obstacles such as walls, trees, or people can block the signal. While 5G networks are being built in low-band (below 2 GHz), mid-band (2-6 GHz), and high-band (above 6 GHz), the distance the signal can travel moves from miles to meters as you progress up the frequencies. As a result, the higher the frequency, the smaller the cell radius, the more cells required to cover an area, the more infrastructure required to support the cells, and the greater the cost of the network.

“While it is not quite that simple, high-band spectrum will play a part in a 5G future—but it is only a part.”

Spectrum policy decisions for 5G represent a trade-off between the physics of the frequencies and the ability to have broad channels capable of carrying large amounts of data. Such broad channels are more readily available in the high spectrum bands. 5G spectrum allocations have thus far focused on these so-called millimeter wave bands. The difficulty with this decision, one anonymous White House official told The Washington Post after President Donald Trump held a White House event to tout an upcoming high-band spectrum auction at the FCC, is that “We are winning a race that no one else is running to build a 5G ecosystem that no one will use.”[17] While it is not quite that simple, high-band spectrum will play a part in a 5G future—but it is only a part.[18]

Low-band spectrum has long been the mainstay of wireless service (and was made even more plentiful by the Obama FCC’s transfer of such spectrum from use by local broadcasters to wireless use). Such low-band spectrum has been assigned to networks in smaller size blocks than high band’s assignment. Therefore, even with aggregation, the total spectrum available to carry the data—and thus the data throughput—is less than high-band.

Mid-band spectrum is “Goldilocks spectrum”—the potential for larger blocks that are not too small for higher throughput, while not too high to impose serious propagation constraints. As a result, mid-band spectrum has increasingly become a focus of policy planners.

However, the potential use of mid-band for 5G runs head long into the other controlling factor in spectrum utilization: the most attractive spectrum has long been allocated for other purposes. These spectrum allocations were made in the analog era, based on the behavior of radio waves. However, the characteristics of digital signals allow more efficient spectrum usage, including spectrum sharing. Your home WiFi, for instance, does not interfere with your neighbor’s because WiFi is a “listen before talk” technology in which packets of data can be inserted into milliseconds of unused capacity.

The “listen before talk” capability of digital technology formed the backbone of the Obama FCC’s “Citizens Broadband Radio Spectrum” (CBRS) allocation. This Goldilocks-zone spectrum between 3.55 and 3.7 MHz is a wide swath, but is being used by the Defense Department, principally the Navy.[19] Using digital signal sensing capabilities, however, it is possible to share the spectrum. A tiering of use priorities was negotiated with the Navy getting priority and commercial users able to plug their packets in when not in use (similar to WiFi, this non-usage is often measured in milliseconds).

Just as no one has repealed the laws of physics, however, so are the laws of human nature in full force. Having received a spectrum assignment, most licensees are loath to relinquish what they consider their God-given spectrum rights to digital performance assumptions like spectrum sharing. As then-Assistant Secretary of Commerce for Technology and Information David Redl told a satellite industry conference: “In this era of competition for spectrum resources, it can be easy to think that we’re in a winner-take-all battle, but that mindset asks us to make false choices that will shortchange America.”[20]

Redl’s remarks were predicated on the on-going reassessment of how to make the so-called C-band spectrum at 3.7 to 4.2 GHz, now licensed for satellite services, available for 5G. Because of this mid-band spectrum’s propagation pattern and the potential for sharing with current users, the C-band holds part of the solution to providing sufficient mid-band spectrum for U.S. 5G networks.

The wireless industry association, CTIA, reports that on average, other countries are making four times as much licensed mid-band spectrum available than the United States. China reportedly plans seven times more mid-band spectrum than in the U.S.[21] In his remarks, Redl addressed the spectrum sharing possibilities in C-band, observing that spectrum policy doesn’t “have to choose between terrestrial 5G and satellite services … These uses are not mutually exclusive; it’s just going to take hard work for them to continue to coexist in a more contentious spectrum environment.”

Network costs are also affected by spectrum choices. As the rest of the world settles in on the mid-band spectrum below 6 GHz for their 5G operations, one can expect there to be scope and scale economies in the production and pricing of mid-band infrastructure and handsets.

No comments: