I saw an article about LTE-U, or LAA LTE in most circles, right before all of the holiday madness. As I read through it and about T-Mobile’s upcoming plans to leverage the unlicensed 5 GHz spectrum for small cell markets, it got me wondering about the spectrum itself, overcrowding in 5 GHz, and how it will affect overall Wi-Fi deployments.
With 802.11ac coming on strong (we’ve recently started on a few outdoor AC installations) it would seem that the 5 GHz spectrum could get really crowded, really quickly. If unlicensed LTE is playing in the same spectrum, at the same power levels, and at the street level, it would seem that this could be yet another set of devices transmitting in a soon to be cluttered area. Has anyone seen any testing of LAA LTE and how it impacts an 802.11a/n/ac network? Scary stuff from just the Wi-Fi side:
|Image Soure: Joey Padden, Lead Architect, CableLabs|
First, let’s recap: what the heck does LAA LTE stand for? It started, it seems, as LTE-U, “LTE Unlicensed”. Then, out of LA-LTE, or Licensed Access LTE, came LAA LTE, or Licensed Assisted Access LTE. Either way, these all refer to the same thing: LTE using unlicensed spectrum in combination with licensed LTE for aggregation and more capacity for devices.
Basically, the LTE groups and 3GPP figured out that there is a pretty large chunk of 5 GHz spectrum out there that can be used to leverage high-speed data and voice, when coupled with traditional LTE. A great write-up of it here and here.
Why you gotta be so rude?
One of the things about LTE, as noted in this article, is that it’s a “rude” technology. Where Wi-Fi devices play nice with CSMA/CD, LTE is, in essence, like the honey badger. That’s not to say that things aren’t being tried to make it a little friendlier, for example Qualcomm proposed CSAT, Carrier Sense Adaptive Transmission, which offers a flexible duty cycle.
So why does playing nicely in the same airspace matter? As if it’s not obvious enough, check out this excerpt from an earlier blog post I came across from CableLabs:
If LTE joins the same channel with a 50% duty cycle, Wi-Fi would now get 50% of the airtime because it would sense the LTE and stop transmitting. In general this means Wi-Fi would get about 50% of the throughput it had in the 100% airtime case.
Half. So if Wi-Fi sees the LTE devices broadcasting, and LTE is using a protocol that sets it to half, then that’s what we have to work with. But half is better than nothing, right? Read the full article here from Joey Padden @ Cable Labs, it’s fantastic!
Well, my concern lies with not only client device connectivity on legacy 802.11 networks, but with networks that are using any type of 5 GHz radios for backhaul.
Take for example a Cambium PTP 650 with auto-rate fallback. A simple way to look at it is, if the link looks good, we can achieve a bunch of throughput. If the link gets worse, the radio drops back in modulation and capacity to keep the link going steady, but at a slower speed. It’s a great feature that a bunch of backhaul radios use. But what happens when that spectrum gets crowded, for half the time? Does that mean that these backhaul links will cycle back and forth if not statically set to hold onto their capacity?
How do you fix it? Well ain’t that the magic question. Follow the TVWS model of a geo-location database? Here is a great post trying to answer that question.
Obviously I’m not the first one to worry about this. There are a few carriers that do support it (TMO, China Telecom, NTT, but you better believe AT&T is a little hesitant with the investments they are making in building out Wi-Fi networks for offload (why would they crowd their precious Wi-Fi spectrum with an unlicensed LTE offering?) I’m guessing Time Warner and Comcast won’t be jumping to help back it either, not that they have any skin in it, but they do have lobbyists who can fight it.
Why am I writing about this now you may ask? Well, there are a bunch of blog posts and articles out there starting in May of last year that go over this. However, at the time no carrier had announced that they were definitely going to move forward with it. Now they have.
It looks as though T-Mobile US (NYSE:TMUS) will be one of the first carriers using License Assisted Access (LAA) in the 5 GHz spectrum band, possibly as early as next year. – FierceWireless
In our industry, it seems like there has always been this split between what the carriers are doing and what everyone else is doing, with the two really never getting in each others respective sandboxes. Well, times they are a changing, friends. First, carriers started to look at Wi-Fi for offload, and now they are looking into spectrum. Granted, it is unlicensed so we all have equal access to it, however we’ve been spoiled enough to think that carriers weren’t interested in the unlicensed side.
Now that T-Mobile has started the trend, who else will jump on board, and how is that going to affect the overall WISP market, the public Wi-Fi market, and everyone else who has been in the unlicensed arena for quite some time? I look to the manufacturers to help hash this out so that we can all play nicely together, but I also keep in mind that the carriers aren’t really worried about how to play nicely if there is a dime on the table to be made.
Update: ISP Radio did a great job with Jack Unger & Matt Larsen regarding this topic a few weeks ago. If you have some time, here’s a link to the site where you can click on “Archive” and grab episode 87.
Update 2: Ericsson breaks news at #CES about leveraging LTE-U indoors. “Ericsson LAA also incorporates fair sharing within the 5 GHz band, to accommodate traditional Wi-Fi users. Fair sharing works on the principle that Wi-Fi and LAA users would have equal access to the spectrum.”
Hi Drew, nice post. One question, you mention that Comcast and Time Warner don't have any skin in the game here… how do you figure? Between the operators included in http://www.cablewifi.com they have one of the nations most ubiquitous Wi-Fi networks (warning: unresearched claim, I just know it's a boatload of APs). Doesn't that count as skin in the game?