Is 5G the answer?

The Story

In 1881 Sleptwell Lindsey introduced a revolutionary new patent medicine which, he promised, cured all problems with digestion, low energy, skin diseases, and sexual performance, the last coded then as “marital happiness.”  He called it “Excalibur” after the magical sword.  He hired the best advertising agency.  He sold more than 4 million bottles before he was arrested.  His nostrum it seemed uniquely combined alcohol and cocaine in proportions that produced euphoria, a kind of desireless state in which one did not care that the medicine cured nothing.  And of course users needed more and more of it.

Lindsey’s ghost is alive today in the world of broadband.  His latest elixir is called 5G.  Its proponents claim that 5G is more transformative than the smart phone or the automobile or electricity.  They attribute this transformation to the Internet of Things, autonomous vehicles, smart cities, Virtual Reality, telemedicine, and factory automation as if only 5G can make them happen.  They intimate network coverage over the next few years with as many as 31 cities already “partly” covered, ignoring the fact that these networks are far from the capacities required for 5G target applications, themselves some distance in the future.  If we set aside the obvious hyperbole of such claims as products of the marketing imagination (although the statements come from corporate presidents), we are still left with a sense that 5G is very important and is materializing soon.  What we may find hard to discern in such statements is the actual proportion of alcohol on the one hand and cocaine on the other, as one contracts and the other expands our sensibilities.  Unhappily, Lindsey himself cannot be reached for an enquiry.

The Current Situation

In its final form 5G is not just a natural evolution of 4G, as 4G was relative to 3G.  It is more of a revolution.  4G raised 3G data rates averaging 5 mbps to an average today of 18 mbps, important for today’s applications, but not a leap into the unknown.  5G touts, and needs for some of its “mission critical” applications, data rates up to 10 gbps (10,000 mbps), a growth of 5000% over 4G.  It also needs round trip delays, called latency, in the order of 1 millisecond (ms), where 4G is now at 30 ms on a good day with no traffic, a reduction of 97%.  Since these figures apply to autonomous vehicles and remote surgeries, they cannot vary with traffic; imagine a mobile network that is not traffic dependent.  At this moment trial “5G” networks provide 300 mbps channels for fixed wireless connections to home WiFi routers, not exactly transformative.  Qualcomm has reported “good results” with a San Francisco simulation (not a trial) that obtained 700 mbps and 7 ms latencies; AT&T recently announced experimental success at 1 gbps.  They are clearly not close.  (By comparison, some businesses and homes in America have 10 gbps services today through fiber optic networks.)  Any sensible look at the state of 5G technology and potential rates of real deployment once the technology has been developed and put into volume production will count the time in many years, even decades for full national coverage.

Consider the Troubles

The obstacles facing rapid deployment of a full-spec 5G network look like one of those medieval castles designed to withstand cannon assault by erecting six-foot thick stone walls.  Here is a partial list: (1) final 5G specifications are not expected from an international consortium until 2020, the holdup largely around how to accommodate 1 million connections in 1 square kilometer, what is thought to be needed for the Internet of Things; (2) no one really knows if its full set of conflicting requirements can be accomplished; (3) it is so complicated that it cannot be fully tested until it fully exists with its applications, which applications cannot be deployed without the network, a classic Catch-22; (4) its cost model is not all that different from wire-line systems when the cost of antennas and Core Networks are included; (5) it requires antennas on poles spaced at a still-to-be-determined distance ranging from 300 feet to 1000 feet from each carrier; (6) each producing radiation that raises concerns in many communities some of whom have outlawed the antennas in residential areas already; (7) each producing less and less reliable signals as their frequencies must increase, and at some point up the frequency scale losing the capacity to penetrate walls and windows;  (8) each connected to the Core Network by new fiber optic cabling; (9) which cabling when installed by three competing carriers will weigh enough to endanger many telephone poles; (10) the installation of which takes from two to four years to accomplish depending upon diverse factors once a city has agreed to it; (11) which antennas will number at least 1 million from each carrier to cover 75% of the country’s population, and could be closer to 10 million, depending upon required density; (12) all targeting applications that do not exist yet or applications with already saturated markets; (13) the new ones each facing serious technical challenges before reaching even emerging market access; and (14) along with the serious, unsolved technical challenges of the network itself.  By the way, none of the four billion smart phones in existence today will talk to a 5G network.  Some 5G compatible phones will appear in 2020, but 5G antennas capable of fulfilling the full promise of 5G have not escaped the lab yet; we cannot be certain that a 5G smart phone sold in 2020 will talk to the 5G network of 2025 at full capacity. In any event, a widely dispersed 5G network does not sound likely for a while, current so-called trials notwithstanding.

But Isn’t 5G In Trials?

AT&T and Verizon have said, many times a day it seems, that they already have 5G networks in “parts of” a number of cities.  Some of these trials are claimed to be with millimeter wave antennas, that is, antennas with radio transceivers operating at frequencies above 30 GHz (the millimeter wave range is 1 to 10 millimeter wave lengths, or 30 to 300 GHz of equivalent frequency—today’s 4G frequencies range from 900 MHz to 1.9 GHz).  Millimeter wave antennas will be required for full 5G compliance.  But the trials are not addressing any of the marquee, transformative applications.  They are serving fixed wireless Internet access at rates in the range of 300 mbps.  That is, they are connecting home WiFi systems to the Internet through a wireless connection, competing head-on with cable television or fiber optic wire-line networks, both of which now can provide services at the same rate or in the case of fiber optics far better.  Verizon’s Sacramento trial offers almost exactly the same consumer package as Comcast and AT&T at the same price, the former already wired to all the homes in the neighborhoods chosen for the trials.  It is not clear what success or failure will prove relative to the technology itself.

The fiber optic cabling installed to connect these trial antennas back to a trial Core Network will likely serve future antennas suited to future applications.  This is a good jump ahead as the fiber cabling is the most expensive part of the future network.  But it is very unlikely that the antennas will serve all future applications.  Judging by what one reads in technical journals on the subject, many serious puzzles remain to be solved before production antennas will be capable of supporting 10 gbps second channels, 1 ms latencies, and as many as 50,000 connections per antenna.  Indeed, until the specification is settled, still a year away, it is hard to see how any fully-compliant antenna can crawl out of development labs, and thence into production, and thence into the field.  Any such process is measured in years.

But Isn’t 5G Cheaper?

The club wielded by 5G advocates over fiber-to-the-home networks is a combination of speed to service and costs; it must be cheaper and quicker to have one antenna serve tens of homes over the air than connect each home to a new fiber optic line, and the air does not need to be broken up and then repaired in areas where a fiber optic line must be routed under streets and sidewalks and through concrete walls into homes.  At face value we must admit the truth of such a claim, indeed the allure of such a claim.  Anyone at Verizon will have horror stories of getting under sidewalks and through walls and up elevator shafts in apartment buildings during its deployment of FIOS.

But the argument, while not fully impeached, does suffer from a few factors.  One is that the antenna and the Core Network are costs not incurred by a fiber-to-the-home network, and they are not insignificant.  A $20,000 antenna (we guess at the price including installation) must actually serve more than 40 homes to break even against the $500 it typically costs to pull a wire from a pole into a home (both must supply home electronics—transceivers and WiFi routers), which will be the most common connection in America with 175 million telephone poles.  At 50% take rate, the antenna must cover 80 homes.  If antennas must be located 300 feet apart, it may take three or four such antennas to actually cover 80 homes, and the nod goes to fiber networks.   There is then the question of each carrier supplying its own antennas (the likely outcome without public policy changes).  Now each antenna is splitting the market into smaller pieces.  If three per cell area, the $60,000 investment starts to look a lot like a losing proposition compared to fiber optics.

Secondly, antennas have active components and mechanical parts with much higher failure rates than passive wiring; any fiber-to-the-home network will be more reliable than any alternative.  Thus recurring costs favor fiber-to-the-home networks.  Over a twenty-year span recurring costs will approach capital costs, hence a figure not to be discounted.  Thirdly, the variability of existing mobile networks is much greater than the variability of wire-line networks.  4G can be as high as 210 mbps and as low as 5 mbps, depending upon many factors but network traffic stands top of the list.  We grant that, in the early phases of 5G, network variability will be much less as 5G capacity is so much greater than 4G.  But when each antenna is trying to talk to 50,000 devices and support autonomous vehicles, smart city traffic conditions, and medical applications that cannot tolerate variations in latency and rate, there would seem to be a time downstream when the perils of 4G creeps into 5G, where the only anodyne is the only anodyne for 4G now—more antennas to create greater network density, each antenna now asked to do less work.

Finally, the killer problem with the argument favoring 5G costs is time.  It is one thing to argue for its superiority as a network in theory, quite another to argue for its superiority when it doesn’t exist.  Having a network with modest speed increases for fixed wireless in parts of fewer than 40 cities, none in the northeast yet, does not a national network with full 5G capacity make.  America has more than 19,000 communities, more than 2.5 million miles of maintained roadway, and 175 million telephone poles.  We are not privy enough to internal design and release schedules to project when the first fully compliant network will be installed somewhere, but the difficulties would make 2025 a sensible guess.  From there to a national network must be a decade, even with many miles of fiber optic cabling installed during early trials.  The current push for fiber-to-the-home networks will have most of America wired by that time.

Isn’t 5G Necessary For All Those Transformative Applications?

Sleptwell Lindsey peddled his nostrum as the only solution to many physical and psychological ailments.  Our mobile carriers have the same line: 5G will make possible huge transformations which, by implication, can only be achieved with a 5G network.  Simply put, this is nonsense.  Autonomous vehicles require a mobile network.  Mobile phones actually in motion require a mobile network, but 75% of all mobile phone traffic today passes through a home or business WiFi network, never reaching a mobile network antenna; it hard to imagine why this would change.  Virtual Reality headsets, rendering their users essentially blind to the world around them, either require cables to a computer or game player, or in more recent models, a WiFi connection, but always with a local computer.  Until latencies are well below 10 ms, remote connections for Virtually Reality are out; but this will come more easily and more quickly from fiber networks, not mobile networks.   Few telemedicine machines require its users to be mobile; most require the opposite.  Many of the objects subject to control in a smart city stand in concrete or lash to poles, no more mobile than the pavement itself.   It is hard to see why objects subject to control within an automated factory need anything beyond WiFi.  We don’t know the percentage, but we suspect most factories of substance are already wired with fiber optics, as are most substantial businesses.

Can’t 5G Use Existing Wiring for Antennas?

The single greatest cost for 5G networks, and a singular impediment to rapid deployment even if all the other technical problems were solved, is fiber optic cabling required to connect antennas back to Core Networks.  The actual amount depends greatly upon antenna density: at 1000 feet between antennas wiring may only need to go down every third or fourth street in one direction; at 300 feet between antennas wiring may have to go down every street in one direction.  In any event, we are talking about hundreds of thousands of miles of cabling.

We already have hundreds of thousands of miles of fiber optic cabling.  While exact figures are hard to find, existing fiber optic networks probably pass as many as a third of American homes and most of America’s businesses.  And that figure grows by the week as more and more communities install their own residential networks, Verizon and AT&T extend their residential networks, and third party providers add to their extensive networks.  Can mobile companies simple connect antennas to these networks and save a lot of time and money in the process?  Can’t future residential and business networks do the same?

In the early going, during what 5G companies loosely describe as “trials,” where antennas are not really asked to do much, the answer may be yes.  But in the end game, when antennas do what 5G claims they will do, the answer is likely no.  Consider that Verizon is installing 5G wire in Boston now with 1728 strands per cable, in an area they already have completely covered with fiber optics for their FIOS system.  Indeed, Verizon boasts that they are installing enough wiring to go to Mars and back, a figure attached to strands, not cabling (if cabling they would girdle the earth five times).  A fat cable these days has 144 strands.   Why is Verizon doing this?  The simplest answer is that Verizon is planning out twenty years. In twenty years we can expect 5G antennas to actually do what is claimed—10 gbps user channels, 1 ms latency, 50,000 connections.  Such an antenna will need back-haul in the 100 gbps range compared to an antenna in trials today that could probably get away with 1 gbps given real traffic profiles and 10 gbps with heavy traffic profiles.  The latter can share lines with other antennas; the former cannot without getting into complex and expensive multiplexors and transceivers, each duplicated at each antenna for reliability.  It seems inescapable that the 5G network when it does provide 5G services will require new cabling for every provider if public policy does not force host neutral antennas, that is, antennas with frequencies for all providers.

Cable Companies Joining the Fray

Recently the two largest cable companies in the country—Comcast and Charter—announced intentions to chase after the 5G crown.  They claim, disingenuously, to have a superior fiber optic infrastructure in place compared to the mobile incumbents.  As noted earlier, such cabling may serve trial applications (which they already served with existing networks), but is unlikely to serve the 5G future, even if it were more plentiful than the infrastructures from Verizon and AT&T, itself doubtful.

Why? Why are all these very large companies betting so much on such a technology that does not exist for applications that do not exist yet?

The Why of It, Now

We have not eavesdropped on executive meetings or board meetings where CEOs ask for billions of dollars to deploy fiber optic cabling in trials and prospectively for real deployment now rather than when the applications are more visible.  But three speculations seem warranted.  (1) They all face saturated markets and have to say something and seem to be doing something to create positions in new growth markets.  (2)  Even though 5G antennas are still in labs and much design work is still needed for Core Networks, the long lead item is fiber optic cabling which of all the pieces is the one that exists in a mature state.  (3) Early revenue opportunities exist related to capturing and reporting user data to retail companies that can be obtained from 5G but not 4G from mobile phone use, an application never mentioned but potentially very important to early sales success.

Market Perception

The market for mobile phones and networks in the United States is saturated; there are more mobile phones than people, and more subscriptions than people; the staggering growth of the last decade is over.  Cable television companies have watched the gradual decline of television subscriptions with “cord cutting” but corresponding growth in Internet access subscriptions for users to obtain IP television, but these markets are also saturated, and the costs for network upgrades are generally not matched by incremental revenues.   It is also likely that we are seeing a pause in the rapid growth of data rate requirements as HD television seems to be more than adequate, other video such as YouTube does not require more bandwidth (and video is now 75% of Internet traffic), and the future bandwidth hogs (Virtual Reality, telemedicine, Internet of Things, holographic video conferencing, 4K and 8K television) seem further away.  So carriers on both sides (mobile and CATV) feel urgency now as both claim to be addressing future growth markets and help all they can to stimulate them, even though so much is out of their control.

Stringing Fiber Optics

The most expensive and the most time consuming aspect of any 5G network arise from fiber optic cabling.  Let us say that Verizon is out for bid on a 5G cabling network covering Hartford.  It will take some months before they have a vendor and a contract.  Then the vendor must conduct an extensive engineering study, for something this large taking six to twelve months. Once the study is finished the vendor must then apply for pole attachment or city permits for any undergrounding.  Pole attachment approval under current conditions can take six months.  Then, and only then, can the vendor start to arrange for pole make-ready, the process of moving cables around on the poles by those who own the cables to make ready for a new one.  In Connecticut that process can take as much as a year with three or four separate companies required to do something that is far from their own commercial interests.  Then cables can be attached to poles or installed in underground conduits, supposing them to have installed during the interval.  This will take a year to eighteen months.  Verizon has now spent three to four years before installing an antenna.  It is confessed in the industry that even small networks with relatively rapid approval processes will still take two years. 5G will happen.  Those with the capital, or a really good story to acquire the capital, have large advantages downstream if they have fiber installed sooner rather than later.

Capturing User Data and Habits

Mobile carriers and cable television companies try to look like they are speeding into the market with instant infrastructures in part to encourage cell phone use over 5G networks.   5G networks will enable the acquisition of information about user locations and habits not available on 4G networks and 4G phones, which information they will sell to product companies, following the model established by Google, Facebook, and Amazon.  This kind of eavesdropping may be repugnant, even scary, but it happens now as we all know, and seems inevitable.  Revenues of course depend upon people buying 5G phones and using them on 5G networks, but such networks can be of the current variety, and even small pockets of real use will have value.

The Sum of it All

5G is not the answer to short-term significant improvement in data access in America; fiber optic networks to homes and businesses is such an answer.  We have no idea how to put Sleptwell Lindsey’s ghost to rest; we expect mobile carriers and cable television companies to keep up the chatter, hoping for induced feelings of euphoria.  But some day the lack of 5G passing most homes in America might temper the PR machines.  Until then, buy water rather than “Excalibur.”