★ THE BLOG ★ Ramblings on WiFi & stuff.

Concerns on LTE-U and Wi-Fi Co-Existance

Estimated reading time: 4 minutes, 22 seconds. Contains 874 words

 

A few days ago Andrew Von Nagy tweeted an article by Patrick Moorhead on Forbes.com, "How LTE-U In Unlicensed Spectrum Helps Carriers Make Money". He was not positive on the article:

After I read the article I replied on Twitter about the apparently one-sided nature of the article and went on to make insinuations of being in the industry's pocket and such. Mr. Moorhead was not amused and replied with his assertion that we were just trolls.

Well, such is the life on the Interwebs. I felt bad and replied assuring him the intent wasn't to troll. We were just venting our frustration that there was no real mention of the concerns regarding LAA-LTE co-existing with Wi-Fi. And that it seems that all we see in the media about LAA-LTE is how it's going to solve all the carriers problems and will even improve everyone's wi-fi. So, after some back and forth - he DM'd me his e-mail and asked me send him my concerns and I took him up on it. I'm thankful the rather than dismissing me he was genuinely interested in my opinion.

So, here is the gist what I put together in the e-mail explaining why as a wireless guy I have serious concerns about the this:


Wi-Fi/802.11 is a "polite", but inefficient protocol. It's half-duplex, only one client can transmit at a time, has inherent overhead that will only be exacerbated by non-wifi interference, etc. Sometimes I'm amazed it works at all! 

Here are the main concerns I have with the introduction of LAA-LTE:

  • Does not play by 802.11 rules.
     
  • LBT (Listen Before Talk/Transmit) is not a given in the US.
     
  • Wi-Fi clients deal with interframe spaces, physical carrier sense (sensing for energy on the medium), Virtual Carrier Sense (NAV timer, based on the duration field in MAC header), and finally let's add a random back off timer to. Unlike 802.11 clients LAA-LTE radios don't know about the timers, random, or otherwise. This can lead to LAA causing delay and decrease in throughput. Carrier sense is not enough.
     
  • In wi-fi, voice and other latency sensitive frames are given weighted opportunity to access the medium. We already have to deal with contention from devices that follow the protocol. Now we add LAA-LTE and we have to deal with devices that don't, and will be deployed on a large scale.
     
  • LTE duty-cycles are an issue as well. CableLabs has done research showing the delay and reduction of throughput cause by varying duty cycles. When duty cycles are small wi-fi clients don't have enough time to contend for the medium. When the duty-cycle increases latency for wi-fi client follows.
     
  • We are already starting to see usage increase in 5GHz and it is difficult enough when 802.11 is contending with itself. Adding another technology that does not follow the rules will only make it more difficult for WLANs to operate effectively. 
     
  • Another issue I wonder about is how will devices choose between LAA-LTE and Wi-Fi? Right now most devices choose wi-fi over cellular for data traffic when available. When LAA and Wi-Fi are co-located how is the decision made?
     
  • Also, why now? More than half of all mobile data is over wi-fi. This will only increase. Where is the benefit for the carriers/users? Because of the power restrictions in 5GHz LAA-LTE will most likely only be deployed using small cells - just like Wi-Fi. So, the most likely places that LAA will be deployed will be areas already served (and probably over-served) by wi-fi. 
     
  • The control plane for LAA-LTE lies in the carriers licensed spectrum where they have full control of the medium. They will now have easy access to unlicensed spectrum for free and also their own licensed spectrum to fall back on. Wi-Fi users don't have that advantage and have nowhere to go. So, LAA starts off with an inherent advantage over wi-fi.
     
  • Healthcare already has a difficult time with wi-fi. Most healthcare WLAN designs do not support DFS channels (UNII-2, UNII-2e) due to the scanning and channel change requirements if RADAR detected. Because of this I'm guessing LAA will not be deployed in those channels. LAA could become even more disruptive in those environments due to the limited number of channels. This, and other dense enterprise deployments may be at risk.

The mobile carriers are looking to grab all the spectrum they can possibly use. There is nothing that will keep them from running over any other users. In the end it's all about making money for the carriers. The same ones that spend billions on licensed spectrum that is theirs alone to use as they please. Nothing wrong with making money, but not at the expense of a useful and expanding technology that is now - even in the nascent stages of 802.11ac - growing rapidly.


04-06-2015 UPDATE:

Tell me again how we can trust the carriers to make LAA-LTE work as a "good neighbor" with Wi-Fi?