Your Network's Edge®
Blog Post
You are here
Can we afford 5G?
Jul 08, 2019
The telecommunications community has been so fixated on the question of whether we can afford not attaining 5G as soon as possible that no-one in the community seems to be asking whether we can afford attaining it.
5G is the first mobile generation that was specifically designed with real-world applications in mind. Hence, many of the technologies being developed for 5G are tightly focused on solving important real-world problems, like vehicle accident prevention (V2V), reducing traffic congestion (V2I), automatic meter reading (AMR), improving accuracy in a variety of tasks using augmented reality (AR), crime reduction (smart city), increasing manufacturing efficiency (IIoT), etc. So, the answer seems to be obvious – the world will benefit from realizing 5G, and the sooner the better.
But there are two caveats to this. First, the focus of 5G deployments (and hype) so far has been purely on increasing data rates for mostly entertainment applications. Second, the focus of 5G research for other applications has not been on finding the optimum solution to real-world problems, but rather on finding a solution to these problems using cellular transmission technologies.
In this blog post I will analyze these two caveats along several axes of affordability and see if the answer to our question changes.
Speed
We are being bombarded with arguments as to why the 100 Mbps provided by 4G is not enough. One popular but unconvincing argument relates to video download time. This argument informs us that downloading a full-length movie using 4G takes about 15 minutes, but with 5G will take only a five seconds (or some similar numbers).
First, let’s check the basis of this argument. 15 minutes at LTE’s nominal 100 Mbps rate translates to downloading over 11 GB, which is quite a big larger than the 3 GB or so of a standard movie (and by the way, 4G is required to deliver 100 Mbps for high mobility UEs, for low mobility users true 4G should actually deliver 1 Gbps!). Perhaps the assumption is that the 4G connection is actually only delivering 30 Mbps or so. On the other hand, the speedup ratio of reducing 15 minutes to say 5 seconds is 180, which is beyond the design goal of between 10 and 100.
More cogently, why in the world would you download an entire movie before viewing it, rather than simply streaming it as you watched? The assumed 4G rate of 30 Mbps is sufficient to stream about five 1 hour 3 GB movies at the same time! Of course, 5G at 1 Gbps could then simultaneously stream about 150 movies, but who would reasonably desire to do that?
Of course, perhaps the real movie-related need for 5G is not for small-screen devices, but rather large Ultra-HD formats. This variant of the argument also falls flat, since the highest resolutions still only require about 25 Mbps - well within the capabilities of 4G.
Another argument that sounds convincing has to do with user generated content. Perhaps 4G is fast enough in the downlink direction, but can’t support users uploading YouTube videos and social network streaming. Sorry, 4G uplink speeds of 50 Mbps suffice to stream 360-degree surround Ultra-HD video plus audio with enough spare for a blog post or two.
The next popular argument is what I facetiously call the proof by induction. This argument states that we really don’t yet know why we will need in excess of a Gbps per end-user, but in the past we have always gobbled up any bandwidth that has been offered. The premise is correct – our requirements in both transmission rates and storage sizes have always risen to whatever technology has delivered. Yet that doesn’t prove the inductive conclusion. It is more likely that technology has not previously reached the ultimate rate that we can consume, but now it has.
The final argument that I will address is that of the ultimate convergence. For many years we have been attempting to converge the manifold information sources into a single unified stream. According to this argument 5G will subsume the telephone, Internet connection, television (cable or DVB-T), home entertainment center, telecommuting conferencing terminal, and connectivity with home IoT, and the kitchen sink. Literally.
The main problem with this argument is not possibility but viability. The fact that the technology enables this ultimate convergence doesn’t mean that it is the most economically viable option. There are, in fact, many impediments, and a host of competitors, along the ultimate convergence road. As a small example, in many jurisdictions there are public broadcast television channels that must be provided free of charge. Will 5G mobile operators provide SIMs and service to non-paying subscribers to conform to this requirement? And what about volume caps? In the 4G world 100 GB per month is considered a reasonable cap, but in this converged world will be consumed in a few seconds. Will 5G mobile operators provide completely flat billing (which for 5G at 10 Gbps rates is effectively a 3.3 Terabyte cap!)? Will consumers be willing to purchase a $100 4TB disk every month to warehouse this data?
Note that I have not mentioned the issue of online gaming. This is not because I don’t believe that this application justifies massive infrastructure expenditures (although I don’t). It’s simply because the major required improvement as compared to 4G is latency, not raw speed. 4G speeds are good enough for all the massively multiplayer games presently foreseen.
Mobile operator bottom line
It is well documented that both CAPEX and OPEX of network operators are proportional to the data rate transported. Mobile operators believe that consumers will be willing to pay a premium for super-fast 5G service, but that premium may be 25% or perhaps even 50% for first adopters. No one believes that a consumer now paying $35 per month, will be willing to pay $350 or $3,500 per month for 5G. In fact, the numbers being discussed are between $50 and $75 for residential customers.
Since mobile operators are already operating on low margins, it is hard to see how 5G is going to be profitable if it is limited to fast Internet access. Yet, right now we are witnessing a propaganda war in which competing service providers are trying to capture the public’s eye with pioneering achievements, while absorbing the losses and waiting for the next, more profitable, phase.
Many mobile operators apparently believe that URLLC services will be their savior. Industry 4.0 may prove to be as world-changing as the industrial revolution, and at very least will facilitate unprecedented levels of automation and inter-site data exchange, for which industrial will be willing to pay. Sometime in 2022 or so mobile operators expect to move into completely new and completely profitable markets and recoup their losses.
Energy
Estimates are that between ½% and 1% of global electric energy consumption is directly attributable to mobile communications. Power consumption can usually be taken to be linearly proportional to data rate (while energy consumption of computation increases super-linearly with clock speed). Hence, 5G’s increasing data rates by a factor of 10 to 100, and increasing its share of the communications market, will directly lead to a major jump in energy consumption and consequently to a steep increase greenhouse gas emission.
Of course, 5G simply must dramatically improve the energy efficiency of the air interface, since a fully charged smartphone battery lasts today about 10 hours, and a power increase of a factor of 50 will reduce that to 20 minutes (not to mention the fact that the phone will become too hot to touch). However, about 20% of the energy use is in the non-air-interface segments, and in addition edge computation is expected to proliferate (and its energy usage increases super-linearly with speed).
ITU-R M.2083.0 put forth the impressive goal of a one-hundred fold reduction in energy consumption per bit in order to precisely counteract the aforementioned issues, but no-one seems to know how this order of magnitude can be obtained. In fact, mmWaves seem to be making the energy efficiencies worse! The mmWave power amplifiers are pitifully inefficient and the copper traces on circuit boards have much higher resistance at these frequencies due to skin effect.
On the other hand, 5G mmWaves do lead to some energy reductions. Since the cells will be smaller the transmit power may be lower (although not that much lower due to more absorption). And smaller cells mean more of the end-to-end path is actually implemented in fiber, rather than in radio.
But if the energy efficiency of fiber is orders of magnitude lower than that of cellular radio, then why not use FTTP instead of 5G FWA? Admittedly, there are cases where fiber access is simply impractical, but we shouldn’t confuse the business plans of a mobile provider without right-of-way with the impossibility of fiber access. In fact, most of the cases which require huge data rates are actually aggregates that can be fed by fiber much more efficiently than by 5G.
Health
Can we afford 5G from the health prospective?
Despite the fact that reports of birds dying en masse from a 5G trial turned out to be a hoax (the birds actually died, but nowhere near the one-day non-mmWave 5G experiment), we can’t dismiss questions of safety, especially when dealing with mmWaves. The World Health Organization’s International Agency for Research on Cancer (IARC) has concluded regarding existing (sub 6GHz) cellular frequencies that “the evidence, while still accumulating, is strong enough to support a conclusion and the 2B (possibly carcinogenic to humans) classification”.
We need to remember that there is simply insufficient data on the long-term safety of continuous mmWave exposure, although it is certain that such radiation has a direct thermal effect on the external 2 mm of skin and eyes. Generally accepted guidelines limit such exposure to 1 mW/cm2 over 30 minutes for RF frequencies in the 1.5-100GHz range, but these guidelines are based solely on thermal effects.
Studies suggest that non-thermal effects of continuous mmWave exposure include (to quote just a few studies) altered structural and functional properties of plasma membranes, skin nerve ending damage, activation of the immune system and peripheral neural system, changes to blood and bone marrow composition, changes to activity of various enzymes, modification of various microbes, increased antibiotic resistance, etc.
We simply don’t know enough to be certain. Proponents of 5G conclude that lack of proof of any direct health risk means that we should rapidly roll out millions of 5G base stations over the next few years. Admittedly this would provide the needed raw data, but at the expense of making all of us (even those without a 5G phone) participate in this grand experiment. Contrariwise, two years ago more than 180 scientists from 36 countries signed an appeal to the EU requesting a moratorium on the roll-out of 5G until potential hazards for human health and the environment have been fully investigated by scientists independent from industry. Since then over 100,000 signatures representing 187 countries have been added to the petition. While it is true that experts on the biological effects of radiation have not participated in the 5G effort, such a complete moratorium is without precedent in the annals of technology. As since it would conceivably take decades to truly study the issue, the requested remedy is overly far-reaching.
Summary
We started this post with the question whether we can afford 5G, and I believe we can now tentatively answer it.
The bad news is that we can’t afford a 5G which is simply 10 to 100 times faster Internet access. Mobile operators literally can’t afford to absorb the huge upfront investments, and shoulder the continuing operational losses for an indefinite period. End-users can’t afford the potentially detrimental effects on health nor the increase in global energy usage and the consequent increase in greenhouse gas emissions. The good news is that we don’t really need to. Most applications don’t really require these speeds, and those which do can be supported by fiber instead.
However, the non-speed related aspects of 5G, having been truly derived to solve real-world applications, are not only required, but are expected to rapidly become indispensable parts of our lives. These applications are the ones that require lower latencies and edge computation, rather than pointlessly high raw speeds. That’s the side of 5G from which mobile operators will profit and end users will benefit (and hence for which they will be willing to pay).
I think I'll wait for an affordable 5G plan.
About RAD's Blog
We’ll be blogging on a wide range of hot topics affecting service providers and critical infrastructure network operators. Our resident experts will be discussing vCPE, Cyber Security, 5G, Industrial IoT and much, much more.