I would expect that LTE will need to provide large and small cells ("metro" cells, as the article calls them), as well as femtocells and/or allow handover to WiFi hot spots. There's no reason why a cellular network cannot take its own architecture down to the fine resolution of WiFi hot spots.
This whole LTE phenomenon is odd to me. It's good that at long last all of the cell carriers appear to have agreed on one standard, but then again, from a purely technical point of view, there's nothing in LTE that wideband CDMA could not have done, or cannot do, equally well.
Why everyone latched on to the LTE wagon, instead of unifying the different 3G CDMA schemes, perhaps taking the best features of each one, is totally beyond me. I have to believe the reasons were financial/IP related, and nothing to do with electrical engineering.
Parenthetically, even the fact that LTE can use RF spectrum slices as small as 1.25 MHz is not unique to LTE. The Qualcomm version of wideband CDMA, cdma2000, can build up channels from equally small slices.