Hurricane LTE-U: Don’t Let Wi-Fi Get Blown Away | Hacker News Reader

Hurricane LTE-U: Don’t Let Wi-Fi Get Blown Away

eff.org

97 pointsDiabloD352 comments

Meanwhile, Qualcomm, which is developing the technology, plied journalists with breakfast and lunch in an effort to convince them LTE-U was totally benign.

As much as I like the idea of a bunch of scheming Qualcomm executives sitting around a giant conference table plotting to ply journalists with breakfast and lunch, maybe it's possible they were just being polite hosts?

Being a polite host is part of convincing people to feel good about you and what you're saying. That's why there is so much entertaining (dinner, drinks, parties, etc.) in politics, sales, and international relations.

There is also at least one study[1] showing that people who have recently eaten have a more positive impression of others.

[1] Mentioned last night on the show "The Brain" on PBS, which I highly recommend.

http://www.pbs.org/the-brain-with-david-eagleman/home/

what fun is advocacy if you can't use connotation to attack your opponents? it's even better when you can make clever puns and show off for your readers. they'll be tickled.

They invited journalists to present exactly this. It's pretty clear. You don't randomly invite journalists just to show off that you can host something.

I can assure you that the "bellhead" v. "IP-Everywhere" battles still exist.

As recently as 2 years ago, international telecom exists were lamenting the proliferation of wifi - arguing that local LTE is more efficient than wifi; and handsets should use local LTE base stations instead of wifi.

Be very afraid of this - telcos will take action to make their delivery mechanisms metered and rated.

"The Rise of the Stupid Network" (1997) is the quintessential essay on this topic, and unfortunately still relevant today. http://www.isen.com/stupid.html

In case anyone wants it, I created an epub of it.

https://r.kyaa.sg/vknoez.epub

LTE probably is more efficient (due to OFDMA vs. CSMA) and I'd be happy for my phone to use femtocells instead of Wi-Fi... as long as the femtos are free and don't stomp on anything. But we shouldn't allow some kind of "protocol Darwinism", especially if it exclusively benefits one particular industry.

I have a very hard time conceiving an at-least-legally-legitimate business as evil as telcos.

Banks, insurance, finance, those generally are perceived as devil spawn, but I really can't imagine another business with such mediocrity, such an ingrained disregard for the customer.

The VA (veteran's affairs)

That's really not a thing in my country (Portugal), although we do have quite a few veterans from our lovely`colonialist wars [1].

[1] https://en.wikipedia.org/wiki/Portuguese_Colonial_War

Indeed, my dad got called a "Circuit switch bigot" by Vint Cerf just a few years ago.

> arguing that local LTE is more efficient than wifi; and handsets should use local LTE base stations instead of wifi.

Interesting, I do exactly this. In parts of my home the Verizon coverage is more reliable than my own wifi.

> In parts of my home the Verizon coverage is more reliable than my own wifi.

This is fixable.

I just wish there was some sort of automated handover for APs on the same subnet. I had the same problem as GP, bu I now have two APs connected to my LAN over powerline, but I have to manually switch them over when I go from one side of the house to the other.

If you set the same SSID, encryption scheme and password on your 2 APs, your devices should roam automatically between them. Though some devices do this more intelligent than others, and if your devices and APs both support 802.11r and 802.11k, you're even better off.

Are 802.11r/k branded somehow? Sometimes spec sheets omit these kind of features. Is there a way to see if the feature's enabled from the NIC's host?

Will this work even if both APs choose the same channel? (In both cases the APs get stronger signals from other networks than from each other, so it's a realistic possibility).

It's your device that roam, it'll chose the AP with the stronger signal, regardless of channel as long as the other parameters match. The APs arn't coordinated, unless you use a controller based setup(which is more an feature of enterprise APs.)

There is: https://en.wikipedia.org/wiki/Wireless_distribution_system

I don't care about wireless interconnection, but wired interconnection.

Of course it is, but when I have great coverage already why bother? Also it can be a pita to fix.

Viewed from this prospective then interference would be a feature, not a bug. Degrading WiFi would result in the use of less WiFi.

Yeah, and users can't even run their own LTE-U base stations because - last I heard - it requires signalling on the licensed LTE frequencies to tell the end device to use the unlicensed LTE-U band for data. So if they make 5.8GHz unusable for anything except LTE-U, users will effectively have to pay the mobile telcos to run their own 5.8GHz access points on their own wired internet connections.

If LTE-U can coexist so peacefully with other unlicensed devices, then maybe all of the LTE bands should be opened to unlicensed devices and carriers can use the same mitigation measures on all LTE bands to let everything coexist happily.

There, problem solved, I'm sure the carriers will jump right on this.

Hm, this is actually a great response. I can't think of a good counter argument. It would be interesting to see someone with more technical or legal experience to comment on the implications of this

I think the counter argument is that the carriers must be automatically cutting back their use of the spectrum when they detect other devices, so if unlicensed devices were allowed on the carrier licensed bands, it would reduce spectrum available to the carriers.

Though it seems fair, if they want to eat into the public unlicensed spectrum, they ought to give the public access to their own spectrum.

If this ever happens around here, I'm replacing my AP with an even louder one, as loud and disruptive as I can go without violating FCC requirements. I will drown out LTE-U in my area and do whatever it take to disable LTE-U detection on the AP (such as buying out of date APs that don't recognize it).

Qualcomm, who is leading the effort behind LTE-U, belongs to the Wi-Fi Alliance and submits things to the IEEE 802.11 Working Group, and its not like they're not making money no matter what they sell.

Maybe Qualcomm should be removed from the Wi-Fi Alliance and banned from the IEEE 802.11 Working Group if they continue to push for standards and "features" (term used loosely) that damage Wi-Fi.

5.8 GHz isn't even good for long distance transmission, this is why it was given away as an unlicensed block to begin with. So, what the hell?

Not only that, cell phone data already has tons of frequency space; they just keep wasting it on 2G, 3G, non-LTE 3.9G, non-VoLTE voice. It is not my problem that "oh, customers are still using hardware that doesn't support more efficient use of the spectrum"; I own a reasonably up to date LTE phone that can do VoLTE, I did not pay out my ass for one (its an original Nexus 5).

LTE is the first protocol that is meant to be fully backwards and forwards compatible, all previous protocols just muck up perfectly good spectrum. You know what happened when USB came out? Or HDMI and DisplayPort? Or PCI-E? Or SATA and SAS? They all quickly replaced all legacy standards and became the norm in short order. Their collective hallmark is easier and cheaper to produce hardware that has a protocol that is forwards and backwards compatible (something they all share with LTE).

LTE, however? In my town, about half the spectrum is dedicated to non-LTE usage, yet most phones around here are LTE data, with or without VoLTE capability, and the few dumb phones remaining are 3G voice.

You know what that means? 5 mbps or less is the norm for downstream data unless you're right in front of the tower, then I've seen it go past 50 mbps. If they shut off 2G completely, shut off 3G/3.9G data, moved all the VoLTE-capable customers over to VoLTE, and shrank the 3G voice segment down to only what we need to serve the area, we'd be getting 10 or 15 mbps instead of 5.

We don't need more cell phone spectrum, we just need to stop wasting it on ancient inefficient protocols.

This is a game you can't win, unfortunately. If the interference gets too much LTE-U phones can just fall back to their parallel LTE connection on licensed spectrum, but your wifi has nowhere to go that LTE can't follow you to.

Single sentence summary for any "authority" doing review on this to make a decision:

"Wi-Fi devices aren’t equipped to recognize the presence of an LTE-U device and don’t know that they should only transmit when the LTE-U device has scheduled itself to remain silent."

and that's why LTE-U cannot be allowed to exist

unless they are just looking for a reason to sell everyone new devices, again

Unfortunately that's not a very good argument on its own. Wifi exists in the unlicensed bands in the US and is thus subject to FCC title 47, part 15 rules. This includes the provision that the onus is on the unlicensed transmitter (in this case your wifi device) to gracefully deal with any interference it might encounter.

At the same time (from the article):

47 U.S. Code § 333 - Willful or malicious interference

No person shall willfully or maliciously interfere with or cause interference to any radio communications of any station licensed or authorized by or under this chapter or operated by the United States Government.

This puts some responsibility on the LTE operator to not stomp on WiFi.

Also, "The FCC has already said that Section 333 applies to unlicensed spectrum by prohibiting Wi-Fi jamming." http://www.wetmachine.com/tales-of-the-sausage-factory/my-in...

While IANAL, the key word there seems to be "malicious". Merely operating in the same band doesn't show any malicious intent, even if it severely degrades performance for other devices in the band.

Jammers on the other hand do intrinsically show malicious intent because they operate in a way which serves no purpose other than denial of service for other local transmitters.

Given that they can't sell these LTE-U devices in Europe/Japan anyway (where rules mandate "listen before talk" like the article says and where vendors have a different standard called LAA), what's the upside in having this just for the US and places like India when experience indicates most manufacturers would rather make one model that they could sell worldwide if feasible? (e.g. almost all GSM phones are quad-band these days and work around the world)

The hardware may be identical and only the software different. They may be able to get away with fewer devices in congested areas if the throughput is better without Listen Before Talk.

Here's my conspiracy theory for the day:

1) In the long-term, regulation is to the benefit of the incumbents.

2) Most in technology prefer light regulation

3) If you act as a "bad citizen" in such a way to piss off people you can change #2.

Thus the result of acting out is more regulation, which increases your costs, but also acts as a significant barrier to entry for future competitors.

Our city and parts of the county have 5.8ghz Ubiquiti equipment serving internet for thousands of residents provided by our local WISP. I also rely on this service myself since I'm just outside of city limits, this would surely be a big problem for everyone. I get a solid 10mbps connection, much better than our only other option 1.5mb DSL.

This is Sherif from Qualcomm. I know LTE-U is a new concept so there's a lot misconceptions around it. I'll be happy to answer your questions.

OK. Will LTE-U devices transmit at the same time as Wi-Fi devices and impair their bandwidth?

LTE-U will first seek out channels with no incumbent Wi-Fi (or LTE-U) access points. If no clear channel is found, it'll seek the least occupied one, then evaluate how many other transmitters are using it. It'll then use the channel for a an amount of time proportional to that total number of transmitters. So if sharing with 2 other Wi-Fi access points, it'll take one third of the time. The result is that the impact of LTE-U on Wi-Fi is no more than the impact of Wi-Fi on Wi-Fi.

I would argue that the impact of LTE-U on Wi-Fi is actually greater than that of Wi-Fi on Wi-Fi, despite what you have pointed out. It is true that LTE-U operates on a fair time-share basis (TDMA [1]), however the problem here is that LTE-U unfairly transmits while Wi-Fi stations are transmitting, giving absolutely no regard to the actual channel usage by Wi-Fi stations. Wi-Fi does not work on a time-share basis, it operates on a "don't talk while I'm talking" basis, and since TDMA is a different way of communicating, Wi-Fi stations don't know how to avoid collisions with LTE-U stations. From the perspective of a Wi-Fi station, an LTE-U station is just like a leaky microwave oven, there is no way to avoid radio collisions with it.

There isn't a way for Wi-Fi stations to know what TDMA time slots are "free," how long those time slots are, and there isn't a way for Wi-Fi networks to "sign up" for a time slot with the TDMA networks. The TDMA networks just decide to talk at some predefined interval without consulting the Wi-Fi networks to see when they would like to talk. This does not sound fair to me - it's kind of like saying "I'm going to yell very loudly once every minute. If it interrupts the conversation that you are trying to have then too bad for you. But I'm not going to yell when my friends are yelling, because we have an agreement in place."

Normally, Wi-Fi stations use CSMA/CA [2] to avoid transmitting when other Wi-Fi stations are transmitting, which overall reduces the number of lost Wi-Fi frames for everybody. This is what I mean by "don't talk while I'm talking." So the impact of Wi-Fi on Wi-Fi is materially lower than the impact of LTE-U on Wi-Fi simply from a statistical standpoint.

In addition, LTE-U only transmits when it detects other Wi-Fi transmitters. However, due to the hidden node problem [3], not all Wi-Fi stations that are using the channel can necessarily be detected by an LTE-U station, but the LTE-U station is still able to interfere with them. So there will be some quiet Wi-Fi stations that the LTE-U station can't see at all, and when the LTE-U station decides the channel is mostly empty and blasts its signal there, the quiet Wi-Fi station will be disproportionately overcome with interference. Again, the interference would be TDMA interference, but the Wi-Fi access points would not know how or when to avoid TDMA collisions, so the interference would be worse than Wi-Fi interference.

Normally, with Wi-Fi and CSMA/CA, the gaps in the louder station's transmissions would be good opportunities for the quieter station to start transmitting, allowing them to sometimes take turns even though the louder station cannot hear the quieter station, and the hidden node problem would be mitigated as best as CSMA/CA is capable of doing so. But with TDMA interference, there isn't any cue that Wi-Fi can take which indicates that now is a good time to transmit without interference.

This could all be avoided if LTE-U didn't solely operate on a TDMA basis, but also ceased transmitting (or skipped time slots) while Wi-Fi networks are transmitting.

LTE-U risks starting or worsening a "loudness war" where other unlicensed spectrum users decide to buy louder and louder unlicensed equipment up to (and possibly past) the legal limit, because their networks are unusable with too much interference otherwise.

Combine this design decision with the interest that wireless telecommunications companies have preferring their LTE networks over the well-being of Wi-Fi networks, and it starts to paint a sinister picture - that telecommunications companies might be attempting to deliberately interfere with unlicensed Wi-Fi networks in order to promote their own networks and equipment as being superior.

[1]: https://en.wikipedia.org/wiki/Time_division_multiple_access [2]: https://en.wikipedia.org/wiki/Carrier_sense_multiple_access_... [3]: https://en.wikipedia.org/wiki/Hidden_node_problem

Sounds bad, but I would like to know if there is a technical benefit excluding Listen Before Talk in some countries. Why would an engineer possibly want to turn it off, given that the support has to be developed for those countries that require Listen Before Talk anyway? (The evil megacorp financial reasons are obvious, but let's pretend for the sake of discussion that they are not stupid enough to think that breaking Wi-Fi networks across the country/world will have no PR repercussions -- it will cause massive hassle for everyone.)

The throughput and latency are probably better if you can just go ahead and transmit when you have something to send.

Right now, HAMs in the US can run personal/private GSM as long as they do it in that little overlapping part of 900mhz band that euro GSM shares with an amateur band...

Do I understand that one could also run a personal/private LTE in this 5.8ghz band ?

Pretty sure the answer is no, because all the signalling to negotiate the use of LTE-U happens over licensed LTE frequencies - so unless you're a ham that happens to be able to transmit in one of those bands, no dice.

Why can't they produce some samples and let the FCC and an independent body of engineers determine if there really is going to be an issue?

Sending out letters with promises doesn't mean squat.

Why not just make it all licensed? I imagine we could solve the coordinatin problem in some fair manner potentially even letting people get payed for not using spectrum.

Establish some kind of protocol for negotiating where in devices acting on behalf of interested parties (you, your neighbors, the local telco) continuously negotiate for spectrum and power. The winner for a slot pay the other parties for their slot.

It's not like violators can hide, so it should be pretty easy to enforce compliance.

This already exists. The modern unlicensed wireless protocols all play nice with each other using some form of carrier-sense.

The problem is that LTE-U is going to use a TDMA scheme, and, according to the article, will not include any collision avoidance. This means that it may start transmitting at the same time your wifi is transmitting.

My point being that a little conflict with my wifi might be ok as long as it's a fair transaction. If I'm freeing the spot for sale and get payed. It would also replace bureaucracy with economics instead, which should be good for optimizing utilization.

The main difference compared to carrier sense is that I'm suggesting a competition based negotiation, not cooperation based. Instead of waiting for a free spot, you buy the spot.

If the unlicensed spectrum is a commons, the problem with uncooperative commoners like this LTE-U thing looks an instance of the tragedy of the commons. In a sense I'm suggesting a form privatization.

I wonder how much it'll be used once this tech start's growing:

http://www.fiercewireless.com/tech/story/fcc-votes-adopt-new...

So if towers start broadcasting on 5.8, won't they have so many more devices to talk to no wifi stands a chance?

Reply on news.ycombinator.com