Gerhard Fettweis gave an interesting talk at the Johannesberg Summit, where he argued that we should abandon the 20-100 GHz range and move into even higher spectrum, above 100 GHz. The reason is mainly cost, above 100 GHz radios and rather advanced antennas can be integrated into the same chip. He also showed propagation results from TU Dresden, illustrating that the radio range can be significant, and that the high-spatial resolution of the antenna allow the resolution of reflections and could potentially allow for spatial/spectrum sharing between users/operators.
Personally, I am excited about the technology and cost factors but I am somewhat skeptical to the indoor use case presented in the talk. The natural application to me is ultra-dense deployment indoor where each access point uses legacy ethernet backhaul (incl power) and would have a range of a few meters. With low cost chips and low cost (existing) backhaul, we would be on the right track to massive capacity. In this setting we should let go and stop worrying about coverage with a single wireless technology). Limited coverage is a blessing that should be embraced as we use walls and obstacles to limit interference and thereby increasing the capacity. A “Wireless friendly” building is a building that lets no radiation out or in, enabling complete reuse of spectrum in the building. When we move outdoor we hand off to another technology more suited for outdoor coverage.
With very dense deployment you can also skip most of the advanced joint processing required for the accurate beamsteering that would require dedicated low-latency backhaul. Instead you would use lower gain antennas and do hand-offs between access points when the user moves or is shadowed. This is something that actually be done over legacy backhaul with only limited delay guarantees.