Forget "5G" marketing hype. LTE is going to a gigabit in 2016 at Telstra in Australia and SK in Korea. AT&T and just about every other large carrier plans 600 megabits to a gigabit in the next few years. (Update September - AT&T confirms 12-18 months.Qualcomm is shipping the gigabit X16 LTE chip. Gigabit LTE uses more bandwidth, 60-100 MHz in total (Carrier aggregation); more antennas, 4x4 and 8x8 MIMO; and more bits per signal, 256 QAM.
Aggregation, MIMO, and 256 QAM will dominate the next five years. 5G highband will be exciting, great pr, and only minimally deployed until at least 2021 and probably a few years later. NTT DOCOMO's CTO Onoe expects nearly all "5G" advances to be under 6 GHz until 2022-2023.
Update Sept: Softbank Japan has turned on world's first 5G. It's Massive MIMO, not mmWave.
Verizon will be deploying some 5G highband to Boston and probably San Francisco very soon, but their CFO Fran Shammo is telling Wall Street not to change their capex forecasts because any large build is years away.
Some of this will inevitably be wrong. Some locations have different needs. But I've learned from many experts who agree on these basics.
Don't take them as always relevant, but I strongly recommend you review closely any proposal that differs.
This note was inspired by some U.N. policy people explaining how important maintaining good 2G and 3G networks will be. I was dismayed because the best in the industry believe the opposite. Verizon and China Mobile have already moved most of their traffic to 4G LTE. AT&T is shutting down 2G in December. Reliance's world-changing network in India is 4G only. 4G phones are only $5-$10 more expensive than 3G, a fraction of the network savings.
This is even more important where wireless will be the only broadband because LTE delivers three to ten times the capacity. The complete shutdown of 2G & 3G will take longer in the developing world because the poor do not upgrade their phones as often. The time to start is now. The Chinese and the Indians are leading the way to everything on 4G.
More spectrum put to use with carrier aggregation. Two dozen countries are delivering 250-450 LTE speeds (shared) by aggregating two or three chunks of 20 MHz spectrum. If you have a good connection to the tower, you will usually get over 100 megabits and often close to the peak. The spectrum crisis was wildly exaggerated. AT&T just told investors they have 40 MHz completely unused, enough to duplicate Verizon's current network. Carrier aggregation makes all the spectrum usable. That's how they will get to a gigabit. Update Sept: DT demoed 1.2gig LTE with 5 carriers.
More antennas (MIMO and MU/Massive MIMO.) Wi-Fi speeds have gone up two to four times in the last few years because advanced routers now ship with three or four antennas. (That's the 3x3 or 4x4 on the box.) Korea's gigabit will use four, fine to a home but too much for most cell phones. Fixed locations can go to eight antennas today. ZTE has demonstrated a 128 antenna system. Your mobile would connect to 2-4 antennas, while other phones would be running on other antennas with less interference (MU MIMO.)
More bits per signal, 256 QAM. The radio frequency analog components are getting better at separating the signal from the noise. Better signals can carry more data.
Wi-Fi first networks. Comcast is ready to turn on 4M home gateways to carry most of the traffic on their coming wireless system. Where wires exist, using more Wi-Fi is by far the cheapest way to add capacity. In rural areas, I believe Wi-Fi mesh is ready. I haven't researched it, however.
Sharing and efficiency "create spectrum." Dave Farber a decade ago showed a chart of spectrum use in New York City. Dennis Roberson has recent data from Chicago. Both showed an enormous amount of unused spectrum even in busy cities. The White House PCAST report reversed previous policy, saying, "The norm for spectrum use should be sharing, not exclusivity." That makes sense, as we all know from Wi-Fi. While PCAST concentrated on government-controlled bandwidth, the same applies throughout. Unused spectrum, particularly in rural areas, should be put to work with a "use it or lose" policy, as recommended by FCC Commissioner Jonathan Adelstein.
Other improvements, from SON to better software and antennas. Many less publicized advances will add up to a major difference. SON - self-organizing networks - rearranges coverage on the fly. For example, the location of cell phone demand changes rapidly when school or work lets out. Cell coverage areas can be rearranged to provide more capacity where it's needed. Beamforming is very effectively increasing performance in Wi-Fi and early 5G highband testing. It works by focusing the signal where it delivers maximum capacity. There are at least half a dozen more proposals, from better antennas to different line coding, whose proponents promise 10-50% improvements. I don't know which will prove out. Some will, also improving things.
Some policy implications.
Sharing is almost always preferable to monopoly use. The telcos need a certain amount of spectrum they completely control, but today's technology allows most spectrum to be shared. Ironically, the telcos themselves are proving this as they test LAA & LTE-U in the Wi-Fi bands. (I believe the current proposals are much too greedy and not in the spirit of fair use.)
It's time to move all traffic to 4G as quickly as possible LTE now delivers 3-10x the throughput of 3G. China Mobile and Verizon have moved most to their traffic to 4G. Almost all phones sold in China and the developed world are 4G. That's starting to happen almost everywhere. 4G phones are only $5-10 more expensive than 3G; they allow the telcos to carry so much more traffic they are happy to eat the premium.
AT&T is turning off their 2G network in December. Verizon has is already "refarming" some of the 3G spectrum for LTE. It's a little harder in the developing world, where fewer people upgrade their phones. They will need several years. But the right move is to get them to 4G as quickly as possibly, to refarm most of the 2G and 3G soon and the rest in the not so distant future.
Spectrum brings down the cost and is crucial where all broadband is wireless. There is plenty of spectrum across the developed world. Most carriers have been cutting capex, something impossible if spectrum really were tight. More spectrum does bring costs down, so it is a good thing. But the impact of changes in spectrum policy rarely is more than 1-2% of the customer bill, far less than marketing.
In the emerging world where landlines are scarce, more spectrum can be crucial to delivering a decent Internet. Wireless networks need to deliver as much capacity as possible. Otherwise, most people will have low caps and won't have the Internet most of us have in the developed world.
There's much more to say. Help me find it.