We had several great articles we wanted to publish at the start of 2018 so we make no apologies for being late with the top trends. With that said and the usual caveat that our ability to predict the hot technologies of 2018 is still about as good as the average election pollster, the CTN staff have yet again put together some hot tips for 2018 and maybe even 2019, in no particular order. In fact what we actually did is resurrect our 2017 list and update, to see how many of them we were more or less right about. There is a comment section at the bottom and alternative or additive suggestions are most welcome.
Alan Gatherer, Editor-in-Chief
5G Sides Safely off the Hype Curve and Makes a Nice Boring Landing
We were a little premature with this one, so we will repeat it again this year. We know of some large deployments this year. In the USA Verizon is aggressively predicting roll out this year and ATT expects to get there before years end. A proper worldwide summary of all of this excitement is an article all by itself. Suffice to say, what is actually meant by 5G is complicated…
Massive MIMO Implementation Proves to Be at Least as Exciting as Massive MIMO Theory
So now we actually want to deploy these things, and for millimeter wave deployments they are essential. But the power and cost remain non trivial issues. There has been a significant uptick in the academic literature on massive MIMO implementation. Searching IEEE explore using “massive MIMO implementation” we found 287 papers since 2005 with 128 in 2017-2018. Single bit ADC solutions seemed to dominate, surrounded by a lot of hybrid pre-coding designs and a few FPGA and even ASIC test beds. More exotic ideas are out there if you look for them. We will continue to see innovation in this space thru 2018 and 2019.
Machine Learning. If You Are Not Doing It, People Will Shun You at Parties
This was the year in which we all decided that a little Machine Learning was good for everything. Egged on by the announcement of Google’s Tensor Processing Unit (TPU), for which the first solid public description emerged in 2017 but is already at v3, suddenly we had Tensor Flow and you didn’t need a CS degree to be a machine learning expert anymore. Silicon Valley’s VC crowd, having gone off the long development cycles of chip design in the early 2000s, got so excited they started funding silicon implementation of AI machines. If you are building a handset SoC and you don’t claim an AI accelerator you will be made to stand facing the corner with a pointy hat on at ISSCC. But we are not too sure what we will do with all of that AI processing yet. Facial recognition? Lets hope there is something more exciting on the horizon. Back in the infrastructure space there are some very serious attempts to apply ML to network optimization. There are even academic proposals to machine learn the modulation on a link by link basis, something we are calling auto tuning. We predict all of this excitement has a couple of year of runway at least.
Internet of Things. Did the Borg Collective Take Over? Well Not Yet.
After our prediction of world domination last year IoT has slid right off the hype curve in our opinion. This despite the fact that the head of Softbank predicted that our shoes would soon by smarter than their wearers. But maybe this has allowed some solid progress to be made quietly by some serious people. Indeed the rate of addition of smart cars and smart watches far outpaced that of new smart phones. Even Amazon was surprised by the popularity of Alexa and last year was the year that ride share bikes really appeared on the streets of large cities. We predict that other unexpected applications will pop up in 2018 and 2019 and the great killer app of IoT is still in our future. Progress has been slowed by what the Economist called the “techlash” against sharing of personal data and the use of drones. A couple of self driving car crashes took the sheen of that technology in the eyes of the public. Significantly, CES didn’t really show any strong new themes, with focus on things like better cellphones, slightly smarter watches, and really big TVs. So much for our prediction of the smart hairbrush…
Driverless Cars, Coasting Down the Hype Curve
OK, so in 2018 we are not all going to work in driverless cars after all. With some of the dazzle taken off Uber and Tesla, as they struggle for various business reasons to make a good profit, we will look more to the traditional car manufacturers to introduce driving enhancement features. But the completion of 5G has given us a framework for V2X (vehicle to everything) communications which will add to our driving experiences once 5G is deployed. DSRC continues to make progress. But even the most basic of driverless applications, truck platooning, has failed to make it to the road yet. So we will back off our frothy predictions and say that solid progress will be made through 2019 in conjunction with 5G deployment.
Security, Privacy and a New Round of Elections
One year into the election meddling scandal, conspiracy, or whatever you may think it is, many people continue to worry about access to private data. As we predicted last year, this may have had a significant impact on enthusiasm about IoT. However it turns out that the biggest security threat isn’t having your information stolen but rather that you gave it away. But we predict that security issues will be a drag factor on IoT technology take up in 2018 and 2019.
Distributed Ledgers, the End of Banks and the Beginning of a New Way to Run Networks
More generally we would like to take some credit for getting this prediction a little right. In the last year there has been a lot of activity in redefining the network as softer, disaggregated and rapidly flexible to new services. All of this aligns with the completion of the 5G standard and we are optimistic that the operators now have the tools in place to rapidly take advantage of new business opportunities using a much more flexible network architecture than we have seen in the past. We predict that the potential and value of this quiet revolution in network architecture will have a large impact on the rate of wireless evolution in the next year or two. But OK, the banks don’t look like they are going anywhere…
There is a folklore in the comms business that all of the even Gs are great and all of the odd ones are terrible. The theory goes that therefore we need to move onto 6G as soon as possible to fix the inevitable issues that the cursed 5G technology will have. Well, we are not so sure, but certainly academia and even some industry labs are already buzzing about what 6G might be. But we figure it is a safe prediction that we will see some activity in the 6G area in 2019. The big question is, will anything interesting come out of it in the next year or so? One very interesting idea that we mentioned above is auto tuning of the modulation itself. This may be a dramatic step but we predict that there will be a continuing trend towards the standard specifying less and less as we leave gaps in the specification for machine learning to set chunks of the modem operation on the fly. Maybe a new era for how standards are developed?
We Are Finally Free from the Tyranny of Moore’s Law
“What?” you say. “We are supposed to like Moore’s law”. But from a design perspective I would argue that it made our lives too easy and also allowed management to treat us poor old design engineers as commodity products. Well, just like it took a couple of year to work out if Voyager has really left the solar system, it has taken a few years for Moore’s Law to properly die, but now it really has. We predict several communication systems related things from this event. First, operators will not be able to expect 2X performance improvement every 2 years. We will have to think harder at what improvements we really care about in the network. Second, communication equipment manufacturers will have to learn to really appreciate their smart design engineers as there is no free lunch any more. Third, communications standards will have to pay attention to implementation a lot more before specifying something that simply isn’t realizable. Fourth, there will be a lot more room for exotic technologies to play a role as commodity CMOS will not smother them anymore. For instance carbon nanotubes, vertical integration, optical compute, advanced packaging, processing in memory, approximate computing, what else? Power brokers in the silicon architecture space are now predicting the rise of “domain specific architectures” and configurable architectures, and though a lot of this may be window dressing on existing technology you can feel the ground start to shift. Basically our jobs are about to get a lot more interesting and that can only be good right?
Finally We Move Past the Cellphone
OK, this probably isn’t going to happen in the next year but we would like it to. Certainly there are some signs of dramatic shifts being made, though as we said above they didn’t make a splash at CES this year. This month a “holographic phone” was announced. We will let you make up your own minds about it but some operators are already promising to offer it. The addition of AI to the cell phone, which we poo pooed above, we will now add to this list as a possible disruptor of the usual functionality of the cell phone. Put just these two together and the opportunity to get out of the screen and allow some intelligent adaptation of the UI to match individual customer satisfaction will be an opportunity to redefine the fundamentals of what a cell phone really is. On top of that there are signs that the firm hold that Android has on the wireless appliance market is loosening due to its lack of penetration in the IoT space. In the opinion of some analysts Android enabled smart phones but perhaps is not enabling their evolution so much. Certainly handset manufacturers need to find more space to compete and differentiate and this pressure will find a way out into new types of devices. So for this year we predict the end of the cell phone and the beginning of a completely new relationship with your wireless device. Something perhaps that will bring together the other 9 trends and give the user a completely new way to appreciate all of this new technology. We can only hope.