Stupid
Aerial designs are fundamentally decided by the wavelength, about 300/(Freq in MHz) metres.
Then gain. Higher gain are more directional once you get past the gain of a 1/4 wave whip. Omni-directional aerials have poor to terrible gain, especially if smaller than 1/4 wave and not in free space.
Frequencies below 900MHz give progressively too much range and poor control of cell size, so poor frequency reuse. The channel size and number of channels is limited. Frequencies above 2.5GHz are progressively poor range, more line of sight and useful only in open plan offices or roof top point to point links.
Speed is related to channel size and signal to noise (power, interference, more aerial gain = more directional). Basically you are very limited in aerial gain (usually negative) due to handset & needed omnidirectional, you are very limited in power. So the signal to noise can't be much different to EDGE on GSM (0.2MHz Channel), or HSPA on 3G (5MHz channnels) or LTE /WiMax on 4G (Up to 20MHz channels). See where the speed is coming from?
The 5G is about integration of bands, infrastructure and logical operators (RAN). Not more speed in the same channel size, nor more bands. More bands are an effort by regulators to make more money and higher ones by Mobile to replace "free" WiFi.
Only the new 2.3GHz and 2.5GHz bands are much use for cellular mobile (3G & 4G are already on 2.1GHz and the old 1.8GHz and 900MHz (0.9GHZ) that used to be GSM only now have 3G and 4G too, depending on country/location.
So can we all ignore the "5G" hype. Almost all of it is totally misleading.
Of course an aerial for a higher band is small!
28GHz is garbage for mobile. Nice for a pair of dishes with LOS, or for up to 10km, a panel array of aerials about 10cm x 10cm x 2cm on your chimney. Totally useless for a handset, except with a pico base-station on the ceiling in each room.
I've a lot of 10GHz terrestrial gear and worked with Ku and Ka band satellite gear. I was working on projects 12 years ago using these bands.