Updates on G.fast, FTTx, WTTR, DOCSIS 3.1, 4G, 5G, and MORE
2017 will be an extraordinary year for new technology in communications. Carriers will be able to double speeds in wireless, cable, and DSL. Multiple networks will roll out dedicated to the Internet of Things.
That said, more than technology is changing. AT&T CEO Randall Stephenson wants AT&T to work like an Internet company. That means substantive changes for AT&T given that Internet companies experiment boldly, find, and fix the problems, and improve their products at a very rapid rate.
It’s been widely published that John Stankey and John Donovan are determined to change 70% of AT&T’s network to software defined networking and virtualization over the next few years. They are the driving force behind the Open Source CORD project: The Central Office Re-architected as a Datacenter. By doing this, their entire world will be managed from the Cloud.
AT&T executives delivered 3 keynotes at a recent open source conference. They are contributing massive amounts of code to the ONOS Project. Suppliers like ADTRAN are getting the message. If they don’t open their code and get deeply involved in SDN, AT&T will be very unhappy.
The new AT&T intends to be able to turn up new services in days or weeks. To keep costs down, almost everything will be designed so the customer can order what they need over the web. As a result of this, AT&T has said they will eliminate tens of thousands of sales and operations jobs over the next few years.
Verizon, Comcast, British Telecom, CenturyLink, Telus, and Windstream, are following the move to SDN, NFV, and open source. They each have slightly different approaches but will probably coalesce around the ideas
of the giants.
Suppliers Follow or Lose
In response, some suppliers will move very quickly to bring new offers to fill that demand. In other cases, the products may not be available for years. Each situation is different and each carrier will set their own investment schedule.
Five or 6 IoT-focused offerings are advancing across the US. Three of these include:
• One such technology advance is low power wide area networks (LPWAN,) such as SigFox. Most IoT applications require very little data, sent intermittently. That allows relatively simple networks to have a long reach with adequate performance. Only one or a handful of receivers can cover all but the largest of cities. Often, unlicensed 900 MHz spectrum is enough. One of these nets, SigFox, plans for 100 US cities by the end of 2016. Altice, owner of Cablevision in the US, is already working with SigFox in Europe. You should be able to find attractive offers. Similar rapid builds are well along in France, Belgium, Germany, and the Netherlands.
• Telco NB-IoT and LTE Cat-M are similar low-volume, low throughput systems using the existing LTE infrastructure and licensed spectrum. AT&T, Vodafone, and Ericsson have announced plans. Seeing the competition coming, AT&T offers aggressive pricing for IoT on its existing network. They charge as little as a few pennies per month for devices that don’t use much bandwidth.
• Satellite is already used for vehicle communication and is targeting IoT in general. Unlike terrestrial networks, satellite can cover remote regions. AT&T works with Globecom on satellite IoT. Inmarsat offers service across Asia.
What Is Enough?
Columbia Professor Eli Noam forecasts 200 megabits will be the minimum acceptable for end users in a few years. The technology to deliver that speed and higher at a reasonable cost is now available on both cable and DSL. And new technologies heading toward higher speeds in the wireless arena will be introduced in the near future. See some of the tactics and their progress below.
DSL: G.fast can now deliver 500 megabits. Fourteen carriers have announced deployments, and 65 more are in testing. British Telecom is committed to 10 million lines of “up to 330 megabits” by 2020. AT&T is enthusiastic. The ITU has approved G.fast Amendments Two and Three, which will help guide providers as they work to deliver speeds over a gigabit.
However, G.fast is limited to short loops: 50 to 300 meters. That’s fine for apartment buildings and dense areas, often served via fiber to the basement. New cabinets are required for those at a longer distance. Still, the cost to do that is less than running fiber all the way home.
Wireless: In 2015, AT&T introduced a new delivery mode: wireless-to-the-rooftop (WTTR.) Gigabit microwave for backhaul costs approximately $3,000-$10,000 to deploy. This is often much less than fiber construction would cost.
More powerful 4-10 gigabit microwave systems are coming Dragonwave and Ericsson. From either the basement or the roof, you can use existing wire to deliver 500 megabits to over a gigabit.
Another notable development is the Google purchase of Webpass in October 2016. The primary business of Webpass has been WTTR for commercial customers. Google now plans to use WTTR for consumer service in San Francisco. That said, they haven’t abandoned fiber, which they continue to expand in some cities.
In terms of a pure mobile play, engineers have been predicting great advances in both 4G and soon 5G for several years. We are now seeing results around the world. Today, dozens of wireless carriers offer peak speeds of 200-400 megabits, using more spectrum (carrier aggregation) and more antennas (MIMO). Telstra in Australia, SK in Korea, and AT&T in the US have announced they will go to a gigabit of 4G LTE in 2017 or 2018. And in September 2016, Deutsche Telekom did a live demonstration of 1.2 gigabits in Berlin.
The first commercial 5G network was turned on at Softbank in Japan in September 2016. They have 128 antennas on each of 100 towers for Massive MIMO. They call it the Giga Monster. In their early tests, they see a 6 to 10 times increase in capacity. Softbank and China Mobile have committed to thousands more.
Top engineers, including Vint Cerf, Andrea Goldsmith, and Arogyaswami Paulraj are optimistic about Massive MIMO delivering a 50x improvement over the next decade. Massive MIMO can be combined with millimeter wave and other advances for even higher capacity.
Verizon’s CEO, Lowell McAdam, will deliver 5G millimeter wave to a limited number of customers in Massachusetts by March 2017. Verizon wants to use mmWave fixed wireless to offer an alternative to cable in non-FiOS areas. There are GHz of spectrum available at high frequencies, enough to deliver speeds of 5 gigabits and more.
McAdam says the range will be 200 to 1,000 feet. That will require a massive network of cells needing expensive backhaul, limiting the near term plans. The testing will provide important information on range and costs. Verizon wants to go full speed ahead on 5G mmWave for fixed wireless, confident they can keep the cost under control. Most other carriers are waiting for mmWave also capable of mobile service, limiting volume until 2021-2023. ABI Research forecasts only 5% of wireless connections will be high frequency in 2026.
Multi-gigabit wireless is ready to emerge from the labs, but with costs unknown most carriers will take time to deploy widely.
Of course, all wireless speeds vary from place to place. Windows and walls can severely reduce performance. Speeds at the cell edge can be 90% lower.
Fiber: AT&T is spending billions to cover 3 million locations with fiber over the next 4 years. Others are also expanding fiber as well. Telefonica in Spain has the widest fiber coverage in Western Europe, with over 80% of Spanish homes passed. Orange in France and Spain is heading toward fifteen million homes. British Telecom, Telecom Italia, and CenturyLink in the US, are adding millions of fiber lines as well, although they are doing more DSL than fiber.
Nearly all new fiber is GPON. It proved it can reliably deliver a gigabit at Google Kansas City. Ten gigabit PON is being deployed by China Telecom in Shanghai.
Cable: The 4 largest cable providers — Comcast, Charter-Time Warner, Cox, and Altice/Cablevision — already offer 200 megabits downstream. My Time Warner Cable in New York is running 200 down and 20 up. RCN offers my partner Jennie a gigabit in a different part of Manhattan. Those speeds have been highly reliable.
Comcast is committed to a gigabit downstream across 45 million homes by the end of 2018. Gigabit is already available in Nashville, Chicago, Detroit, and Atlanta. Denver, Colorado; Indianapolis, Indiana; Jacksonville, Florida; Kansas City, Missouri; Knoxville, Tennessee; Portland, Oregon; San Francisco Bay Area, California; San Jose, California; Salt Lake City, Utah; and Seattle, Washington are scheduled for early 2017.
Comcast’s Jorge Salinger predicts they will achieve hundreds of megabits of DOCSIS 3.1 bonded upstream in 2017. Casa Systems already is demonstrating 400 megabits upstream at trade shows. Many CMTS systems in the field will only require a software upgrade.
Interestingly, CableLabs is currently involved in an active project for Full Duplex (FD) cable. FD uses the same spectrum for downstream and upstream. When ready in a few years, Full Duplex is expected to raise upstream speeds over 500 megabits and possibly to a gigabit.