5G Flavors: mmWave is good, low & mid-band closer to 4G

Wireless has been getting better at a ferocious rate since at least 2013. Verizon calculates its cost per bit has been going down 40%-50 every year and expects that to continue. This has been true and is a good thing. “5G” has become a meaningless marketing term; most of the big advances happened in the 4G standard. (Massive MIMO, Carrier aggregation) Wireless keeps getting better, whatever the name or the “G.” 

Millimeter wave does get to ~ 3 times the speed because there’s a lot of empty spectrum up there. It has more capacity, so the carrier can offer much more bandwidth.

No 5G has very low latency today, contrary to many claims. Latency in 4G is 25-60 ms, mostly 40-55 ms in the U.S. The average 4G latency in Singapore is 29 ms. 5G today has little effect on latency.

Verizon officially states their 5G is 30 ms although some people have found lower in tests. “Air latency,” from your phone to the tower, is irrelevant for most purposes because servers are not at the tower. The time from the tower through the network is included in almost all tests of latency. This varies enormously as Verizon and others drastically reduce backhaul delay.

The 1 ms often cited only works in the lab. Factory and similar small 5G systems can get down to 10-15 ms because they have no backhaul.

Some locations will see latency drop to 10-20 ms – in 4G & 5G – as new servers are placed at the “Edge.” This will be widespread in a few years in China, Rakuten in Japan, and Jio in India. Verizon is deploying 50 regional Edge servers and possibly more in time.

5G URLLC (Ultra-reliable low latency connection) one day will cut about 5 ms off either figure when it comes out of the laboratories. The gain is mostly from a software upgrade (Short TTI,) which is included in the 4G standard as well.

All governments lie, and corporate lobbyists are mostly people who have left government and lowered their standards. Some get paid millions and are able to convince people 2+2=5, especially in D.C.

Millimeter wave: 23 GHz and above. 

That’s the one that really does get gigabits to most. That will probably get to about a third of the USA, almost all on Verizon and most not soon. Shorter wavelengths have less reach, so they need many more cell sites, which costs money. Verizon decided late in 2018 not to offer that to all. 

T-Mobile and AT&T did some mmWave early but that was mostly pr. Conveniently, T-Mobile provided a map that gives a nice illustration.

Millimeter wave has more capacity because there’s a lot of spectrum up there. Verizon has 800 MHz, about 6 times what it has in the traditional bands.

Mid-band: 2.5 GHz to 4.2 GHz 

This is actually good stuff, whether in 4G or 5G. With Massive MIMO, it has decent reach so is much cheaper than mmWave. 90+% of decent 5G in the world is mid-band.

Mid-band mostly is deployed by upgrading existing sites. It’s much cheaper than adding new sites, either small cells or towers. Mid & low-band 5G by costs carrier the same or less than 4G. The “High cost of 5G” was always exaggerated and only applied to the rarely provided mmWave. It’s amazing how many people believe old claims that have nothing to do with how 

T-Mobile has 160 MHz at 2.5 GHz, more than twice what it has in lower bands. The satellite guys had about 400 MHz, most of which they didn’t need. Few cared, because until Tom Marzetta invented Massive MIMO at Bell Labs in 2010, these frequencies didn’t work well for terrestrial data.

(Massive MIMO uses lots of small antennas, usually 64, and typically improves performance 3-4X. Results vary. That’s enough to make mid-band spectrum economical. It was first deployed by ZTE and Huawei in China and Japan in 2017 and reached the U.S. in 2019. It was developed for 4G, works in 5G, and is the main reason for high performance in 5G.)

Mid-band has enough spectrum to deliver 100-400 megabits to most users, so is the golden band of wireless. In the lab speeds in 4G & 5G top a gigabit, but that’s rare in the field. “Up to” speeds are b_____ if you don’t get them in the real world.

Low-band: 450 MHz to 2.1 GHz

Most engineers did not consider this range “5G” because the speeds are similar to “4G.” The CEO of BT and top technical people of DT, FT, and Telefonica late in 2017 spoke of 5G as microwave only. Wikipedia pointed out the ambiguity.

In a remarkable pr coup, the company group, 3GPP, in 2018 declared almost all new networks “5G” if they included the “5G NR” software. This confused almost everyone; suddenly, what would have been called 4G was now “5G.”

The hype around 5G was already ridiculous. People thought “5G” was wonderful even if they didn’t know what it was. Since “5G NR” software does almost nothing at frequencies of 2.1 GHz or below, this is ridiculous.

Because the marketing folks and lobbyists were so persuasive, the world now thinks anything with “NR” software is “5G,” even when it is is slower than decent 4G.

4G in production at T-Mobile in Manhattan tested over 500 megabits down in 2019. It used carrier aggregation and LAA, which put part of the signal in the unlicensed Wi-Fi band. All 4G was headed toward similar.

The “low-band 5G” of AT&T and T-Mobile is more likely 50-75 Mbps because carrier aggregation and LAA aren’t quite ready for 5G. When they are, “low-band 5G” should be the same speed as 4G or perhaps 10-20% faster.

Real 5G and fake 5G

Promotors, corporate and government, are comfortable calling all of the above “5G,” the 3GPP decision. I think “5G” should only be applied to higher speeds than 4G, as did most engineers. We’ve been steamrollered by the hype of a trillion-dollar industry.

For my analysis of 5G numbers, I try to exclude low-band. The companies rarely separate the figures.

Talk to me

Scroll to top