Massive MIMO is the next great step in mobile networks. Huang Yuhong of China Mobile and Hidebumi Kitahara of SoftBank are deploying hundreds of nodes and reporting performance improvements of 2X to 10X. Sprint in the U.S. and 3 in Austria are firmly committed. Gig LTE - 4x4 MIMO, 3/4 band aggregation, and 256 QAM is spreading rapidly in 2017; the next step will be the 64+ antennas and beamforming of Massive MIMO, some at millimeter wave. I had a chance to ask some experts a few questions so put together these. Others welcome.
in 2018-2019, what's a smart Massive MIMO deployment strategy for a network with growing traffic?
At least 4x4 is the minimum for anyone building today, with many considering massive. For those with TDD networks, is there any reason today not to go massive where you need the capacity? If you don't need the capacity today but want an easy upgrade path later, what's the right design? (Are extra antennas cheap enough to go ahead; differences in radios; ...)
Can analog antennas bring down the cost, especially in upgrades? (AT&T thinks so)
What kind of terrain is likely to be on the low end (?3X) gain and where can you expect better?
All the Massive MIMO deployments have been single band, ~20 MHz. Can we aggregate three or four bands if they are contiguous? (Mexico) If the bands are not contiguous, when will that become practical?
M-MIMO TDD-LTE is deploying at China Mobile & SoftBank, but other carriers need FDD-LTE. How soon can we expect widely deployed quantities of FDD-LTE?
SoftBank says they have tuned their network for better performance at the system edge. What kind of flexibility do system architects have?
What other questions should system architects be asking?
In 2020-2024, what should we expect?
Massive-MIMO is still developing, with many software improvements likely. If today we are getting 3X-10X, how much better are we likely to do?
What should we be expecting in improved antennas?
Massive MIMO antennas are today too large for small cells except in millimeter wave. Will massive MIMO become practical for small cells in bands below 2 GHz? Bands below 5 GHz?
What's realistic for the low end of latency for Massive MIMO in a midband, say 3.5 GHz?