Why the future is not super-fast 5G

Will global wireless networks change as much in the period 2015 – 2025 as they did in 2005 – 2015?

Professor William Webb

Introduction

In 2000 I wrote a book entitled “The Future of Wireless Communications”. I made predictions for 2005, 2010, 2015 and 2020. My predictions for 2015 have proven to be almost spot-on. While I did not predict the iPhone I was clear that the phone would tend towards a “remote control on life”, able to help us in all aspects of our daily routines. I was correct about the data rates available in the home and while mobile and in the overall form of our communications. My predictions for 2020 were that little more would change – once we had achieved the 2015 vision then there was not much more that consumers would need or could be economically provided.

Yet we stand on the brink of the design of 5G mobile communications, a solution that many see as being revolutionary, promising a thousand-fold increase in data rates and capacity, huge reductions in delay and all for a lower cost and energy usage than today’s solutions. Was my vision wrong? Will the world change as much, or more, in the period 2015 – 2025 as it did in 2005 – 2015? I believe many of the 5G premises are flawed and that we will tend towards a better remote control on life (RCOL) than a faster phone.

The last ten years

Between 2005 and 2015 we saw an incredible change in mobile communications. Much of this was driven by the introduction of the iPhone in 2007. This provided a new way of interacting with the device and the powerful concept of the Apps Store. The iPhone heralded an era of smartphones of increasing power and size as well as a dramatic increase in the usage of mobile data. In practice, this dramatic change came about because the iPhone was the last piece in the puzzle – with other pieces already in place including the Internet, mobile data networks able to deliver relatively high speeds, powerful processors, touchscreens and the iTunes store that allowed downloads and synchronisation. These had been developed over many years – so the changes that consumers saw between 2005 and 2010 were really based on underlying developments over the previous two decades and longer.

But since 2010 much less has changed. The iPhone has reached its 6th generation but is still much the same as the 1st, just a bit slimmer, bigger and with more memory and processing power. Similarly for the iPad and laptop computers. It is telling that the best that Samsung could do to differentiate the S6 from the S5 was to curve the screen around the edge, although it is unclear what problem this solves. The world has been introduced to wearables, but Google Glass was withdrawn and fitness devices have failed to become ubiquitous. The jury is still out as to whether the smart watch will be popular and whether it will change our lives in the way that the iPhone did. It seems unlikely.

The next generation network – 4G – has also arrived. Much is claimed but in practice it offers speeds only around twice that available via the latest 3G systems. This is a welcome increase but hardly a major one. One of the main needs for speed seems to be the need to keep all the Apps updated!

A trend that has been creeping up on us is the ability for the phone to control our lives. Systems like Google Now and Apple’s Siri can read calendars, estimate journey times and set alarms. They can provide intelligent reminders and answer factual questions. They are starting to interact with intelligent home thermostats such as NEST to control heating systems automatically based on calendar entries and weather predictions. There is still a very long way to go until they are as good as a well-qualified PA, but with ever more big data and pattern recognition and more “Internet of Things” connected devices there is little reason why they should not get there. The RCOL is not dependent on the handset or the network – it uses cloud storage and big-data cloud processing to derive its intelligence. Indeed, it is in this “over the top” area of services, social media and more that we have seen the change in the 2010-2015 era, building on the smartphone platform and increasingly standardised operating systems.

The next ten years

Attention is now turning to 5G. And the assumption is broadly that it will be an extrapolation of previous generations as shown below.

 

Broadly, 5G is expected to be 10 times faster than 4G (some say 100 or 1000 times), with 10 times lower latency, all while keeping cost and energy consumption to similar levels to 4G. Extrapolation is a powerful tool – for example Moore’s Law has held for many decades and provided a very sound basis for predicting across multiple industries. But all such empirical laws break down eventually, the question is just when. Is it time to seek a different way of looking at the next generation of mobile phone that has been used in the past? I believe it is for multiple reasons:

  1. There is little need for higher speed.
  2. There is strong need for other improvements such as IoT and reliability.
  3. The cost and complexity of achieving higher speeds are growing much faster than the benefits.

Let us look at each of these.

Firstly speed. 4G solutions deliver around 20Mbits/s with enhancements expected. For 80% or more of the time phones are connected via Wi-Fi which is able to deliver 100Mbits/s, again with enhancements expected. Watching video on a small phone screen only requires about 1Mbits/s and downloading a large file of say 10Mbytes is already possible in 1s – would reducing that to 0.1s make a material difference? Web pages might load very marginally faster but the benefit of this will be small. That is not to say that faster speed or lower latency is undesirable – far from it. But that they may not be valued particularly highly over what is going to be available by the time 5G is introduced.

Secondly, in delivering the RCOL, there is a need for huge growth in the Internet of Things. Some have connected thermostats but not yet many other home appliances, parking space sensors, trashcans, doors, cars and so much more. Connecting our world could lead to a major change in our ability to sense and control our lives. This is a much more important area of focus than speed. Integration of IoT into the cellular environment is envisaged as part of 5G and this is one element that I support.

Thirdly, even if speed were needed it is far from clear we can provide it. There are a number of issues here:

  • Handset processing power.
  • Handset batteries.
  • Radio spectrum and associated radio access technologies.
  • Economics and network ownership issues.

Again, let us run through these. Firstly, if we see a 10-fold increase in data rates then we will, at least, need a 10-fold increase in device processing power (often more than 10-fold as more complexity is needed to achieve more throughput). But, as the chart below shows, device processing power might not be able to grow for much longer.

 

The chart shows that clock speed of the typical processor of the era (not necessarily the one in a phone) grew quickly between 1G and 3G, outpacing the growth in data rate. But since around 2002 clock rates have stalled. By 4G the faster than needed growth between 1G – 3G had been eaten up and unless something changes in chip technology it will be very difficult to cost-effectively design handsets able to handle 5G data rates. Multiple cores and dedicated accelerators might help but that then leads to the second issue of battery life. Batteries have only improved slowly and today’s phones struggle to last a day whereas the best 2G phones easily managed a week. Unless there is a major advance, our 5G phones will manage a few hours at best when used to their potential. Even if there is an advance in battery technology able to overcome the extra power required, the energy dissipated will mean we will need heat-insulating gloves to hold the phone as it runs vast computing resources.

But let us assume all these problems can miraculously be solved. The next issue is radio spectrum. Higher data rates require more bandwidth. A 10x increase requires around the same increase in spectrum availability. This much spectrum is only available in frequency bands above 20GHz – well above any bands used for mobile communications to date. In these bands, effective communications requires narrow beams of energy steered to mobiles which have complex arrays of antennas to receive and signal back. The handset will be much more complex and the difficulties in tracking mobiles effectively with narrow beams of energy are immense – to date there are no effective solutions to this. Then we need to install hundreds of thousands if not millions of these cells since the maximum range will be only slightly greater than Wi-Fi. Which does beg the question as to why not just use Wi-Fi?

Finally, it is worth noting that throughout the recent period of rapid data growth, the amount we pay per month for service (excluding the phone subsidy) has generally gone down. While we like higher speeds we are not prepared to pay more for them. Indeed, one of the big attractions of 4G was not for the consumer, but for the operator which could reduce its cost of data provision through implementing a solution designed for the mobile data era. Few expect to see consumer price rises for 5G data rates. Perhaps the operator that does not implement 5G and so provides a lower-cost service to consumers will do better than the leading-edge service others strive to provide. That has certainly been the trend in many other areas of our lives from air travel to clothing.

Is there a better alternative? I believe we should be heading for a better RCOL rather than a faster phone. This would bring real value to the consumer and is actually the direction we have been headed for at least the last five years. A better RCOL would require:

  • Widespread IoT solutions in the home, office and wider area.
  • Big data analytics that can learn from a population and apply this along with artificial intelligence to guide individuals.
  • Changes to systems such as those in shops to recognise individuals and provide them with tailored services.
  • Ubiquitous coverage and totally reliable minimum data rates such that the RCOL can always function optimally.

For this we do not need 5G, at least not the high speed elements that are currently at the core of what 5G will be. We do need a ubiquitous standard for IoT connectivity and a technology that allows wide-reaching reliable communications likely using a simpler air interface than 5G envisages in frequencies below 1GHz.

In summary

Mobile communications has realised much of the promise it offered back in 2000. The handset is heading towards becoming a remote control on life, providing extraordinary value to individuals. Through deploying IoT, harnessing big data and ensuring excellent coverage and guaranteed minimum data throughput levels this vision could be realised. However, 5G threatens to take us in a different direction – away from simple device connectivity to ultra-complex solutions enabling data rates well beyond those of any significant rise in consumer value. As 5G research and standardisation unfolds the problems this will cause us will become ever more apparent. It is time to stop extrapolating data rates and take a more intelligent view of the future.

Biography

William Webb is the Mobile Communications series editor for Artech House Books; is CEO of the Weightless SIG, the standards body developing a new global M2M technology; is a Director at Webb Search, an independent consultancy; and is President of the IET – Europe’s largest Professional Engineering body. He can be contacted at wwebb@theiet.org.