In 2014, Stanford Professor Paulraj told me MU MIMO would be the most effective way to deliver the capacity Africa needed. In 2016, the first Massive MIMO systems were deployed by Softbank Japan and China Mobile. The results were excellent. Softbank reported “up to a 10X improvement.” Huang Yuhong of China Mobile told me 3X was more typical, which she considered an exceptional result.

This note is over-simplified, of course. Wireless networks are extremely complicated and results vary enormously with terrain. I wanted to write something because both academics and network builders have told me Massive MIMO, 4G or 5G, should almost always be on the table planning networks were wires are few.

Massive MIMO also can lead to the effective doubling of primary wireless spectrum. Without MIMO, 2.5 GHz spectrum was in limited use. Midband 3.3GHz to 4.2 GHz was impractical for wireless networks. With the greater performance of Massive MIMO, 2.5 GHz has become golden spectrum that is delivering America’s best network at T-Mobile. 90%+ of 5G worldwide is midband, with superior speeds.

Africa needs capacity, not raw speed. Very few applications need more than 25 Mbps and most work fine in less than 10 Mbps. Speed, however, is a good proxy for capacity. A system that delivers 200 Mbps would not actually support 50 4 Mbps streams, but that’s a useful way to think about things. Few users will actually draw 4 Mbps; most video and almost all web surfing run at lower rates. With statistical multiplexing, as many as 1,000 phones could be supported.

Africa’s Internet is nearly all wireless because there are very few landlines. The result: Most African Internet connections are wireless and have low caps. The average African uses 90% fewer gigabytes each month than Europeans with landlines. Most Africans can’t afford to watch much video.

04 December 2016

“I am crazy about Massive MIMO,” Kitihara of Softbank ordering 1,000’s of Massive MIMO bases

China Mobile and Softbank plan thousands of sites. At Huawei’s remarkable Tokyo Mobile BBF, both Softbank Japan and China Mobile showed enthusiasm. These are the first large moves in 5G, pulling Massive MIMO at least one year ahead of high-frequency millimeter wave. I believe my report Softbank’s Giga Monster Massive MIMO: World’s first commercial 5G is not highband millimeter wave was the first in the West. I have since learned that Huawei, as well as ZTE, provided equipment.

Gear that triples capacity (or better) at modest cost is now in production from both Huawei and ZTE. Softbank has 100 base stations up and running. They are mostly in Tokyo but also in several other cities. China Mobile has working systems in two cities.  

Their early reports are that 128 antenna Massive MIMO today increases capacity 3X to 10X. Expect considerably better results over time. MIMO inventor Paulraj of Stanford two decades ago realized that MIMO could enormously increase capacity in a given amount of spectrum. Remarkably, the additional throughput requires minimal additional power.

Respected engineers – Henry Samueli, Andrea Goldsmith, Vint Cerf – two years ago were comfortable with predictions that wireless capacity would soon increase 50X. MIMO is crucial to that goal.

The results of China Mobile’s testing were also strong. Huang Yuhong, their 5G lead, tells me they will go into production deployments in 2017. She expects they will move rapidly to volume. China Mobile already has a dense network, with 1.3M bases for their 850M users. That’s about 650 users per base station. As subscribers demand more, they need to continue adding capacity.

Robert Clark of Light Reading reports Kitahara intends to use the capacity to improve performance for all users. “We don’t care about peak throughput. We want to minimize the low throughput, which is between 1 Mbit/s and 5 Mbit/s.” Clark adds, “At those speeds, the network will struggle to deliver a good video experience. Prior to deployment, around 20% of all users were experiencing throughput below 2 Mbit/s.” Kitahara reveals, “only a few percent,” now receive low speeds. The massive MIMO solution, using 128 transmitters, ensured that users were getting at least 5 Mbit/s data throughput. Kitahara’s comment, “I am crazy about Massive MIMO,” is also from LR

(My conclusion China Mobile’s network is dense comes mostly from a comparison with Verizon and AT&T. They each have between 50,000 and 70,000 bases for well over 100M subscribers each. That’s over 1,500 users per cell site. Because of different ways to count small cells and DAS, all comparisons here are imprecise. The gap between the U.S. and China is large enough to make the comparison valid. Robin Bienenstock of Bernstein did a similar comparison between California and Spain and found a similar gap. From Barack Obama to the head of CableLabs, Americans continually assert the U.S. has a great wireless network. They are mistaken; U.S. networks don’t match China, Japan, or most of Western Europe.)  

 Tom Marzetta of Bell Labs is credited with the invention of Massive MIMO. See my report, Tom Marzetta’s work on Massive MIMO. He expects:

• Orders of magnitude spectral efficiency gains over LTE – large numbers of users communicate simultaneously over entire allotted spectrum through elementary multiplexing signal processing

• Uniformly excellent service throughout the cell – regardless of location relative to base station

• Drastically reduced radiated power

• Simple and scalable design – employs measured channel characteristics rather than assumed channel characteristics