Hacker News new | past | comments | ask | show | jobs | submit login
Facebook reveals 'SuperCell' tower to increase connectivity in rural areas (fb.com)
134 points by mostdefinite1 on Dec 22, 2020 | hide | past | favorite | 95 comments



Lost interest when I read:

>The test location had a rocky desert topology with minimal foliage, and tests were conducted using LTE Band 41 (2,500 MHz).

T-Mobile/Sprint use this band. I have a lot of experience with it because I am located in a semi-rural and hilly area and I use LTE exclusively for internet access.

For B41, I have one tower that's approx 2 miles away and 600ft or so down from me in a valley, close to line-of-sight. This signal is only usable if I install an antenna outdoors or use next to a window facing the tower. My real-world speedtests are around 75mbit down, and 2mbit up on a good day. Indoors not in front of a window I cannot force any of the devices I own to connect to this band.

I'm also going to mention: Uplink/transmit power on most cellular devices is limited to 200mW (class 1) unless you use a device that supports Sprint's Class 2 High Performance User Equipment (HPUE) at 400mW or you have a router/device that does class 3 High Power User Equipment (also HPUE) at 1.25W. Only class 3 devices have a shot in hell at communicating back to this tower if there's even a small number of trees or a single wall between you and it.

Finally, this B41 TD-LTE tends to work worse the further you get outside the cell, since it relies on precise timing of all equipment inside it. Installing cell relay equipment works poorly outside of a mile or so of the transmitting antenna, again due to precise timing issues. (personal experience with a Sprint Magic Box).

Ignoring all the other social/anticompetitive issues there is with Facebook proposing this, it stinks like a scam to get a tax writeoff or jump onto the same sort of FCC rural internet grant money Starlink is bilking taxpayers out of. You can't scale a midband macro like this even using a dumb amount of spectrum. Oh totally forgot, there's an 'educational' use case for use of BRS (2500mhz) spectrum that I'll bet Facebook will also try to get their hands on.

EDIT: Note on power class, you have certify through the FCC initially to use higher classes, can't just enable it later on without recertification. Also, higher transmit power = less battery life. This is why in urban areas it usually makes more sense to use cell network to save battery (better scheduling/sleep modes, low transmit power) than it does in rural areas where the cell radio is cranked up to its limits and it makes more sense to use a low power radio like wifi broadcasting all the time.


> Lost interest when I read: > >The test location had a rocky desert topology with minimal foliage, and tests were conducted using LTE Band 41 (2,500 MHz).

Testing was conducted in other locations as well. One use-case being discussed doesn't exclude the others.

For example:

> The Metcalf location has flat terrain with thick deciduous trees covering the land.

Also...

> Ignoring all the other social/anticompetitive issues there is with Facebook proposing this, it stinks like a scam to get a tax writeoff or jump onto the same sort of FCC rural internet grant money Starlink is bilking taxpayers out of.

Facebook doesn't build and operate this kind of network. This project (and others such as efforts in middle mile fiber construction and funding cable consortiums) are largely intended to help regional carriers outside of the US past hurdles around capacity and density that limit their growth. Since it's not targeted at the US, the FCC's grants, spectrum allocation, and power limits are not relevant.

Cui bono?

Facebook's learned that just getting people connected to the internet in a content-agnostic way is sufficient for growth. This is why FB funds consortium fiber around and across Africa, funds internet exchanges, and makes free network planning tools available to carriers.

https://www.internetsociety.org/issues/ixps/facebook-ixp-par...

Disclaimer: I'm a Network PE at Facebook, in prod, not in connectivity labs where this project originated, and def not speaking for the company.


> Testing was conducted in other locations as well. One use-case being discussed doesn't exclude the others.

Except the comment you're responding to doesn't seem to take issue with the location. It goes on at some length about the problems with LTE Band 41.

How does pointing out that testing occurred in other locations address the problems with Band 41?


>Ignoring all the other social/anticompetitive issues there is with Facebook proposing this, it stinks like a scam to get a tax writeoff or jump onto the same sort of FCC rural internet grant money Starlink is bilking taxpayers out of. You can't scale a midband macro like this even using a dumb amount of spectrum. Oh totally forgot, there's an 'educational' use case for use of BRS (2500mhz) spectrum that I'll bet Facebook will also try to get their hands on.

Is it very clear that this pessimism/cynicism is warranted?

https://www.youtube.com/watch?v=XzXZdipxCuw


>Is it very clear that this pessimism/cynicism is warranted?

Yes, read some of my earlier comments regarding Starlink, DISH, and general telecom schemes. I'm not coming out of nowhere with this criticism.

With that being said I fully understand where their system is better than current MEO/GEO systems for low-latency rural deploys. I'm just not in the cult of Elon, physics/money/practicality have their limits.

I'd honestly be more excited with Starlink if they were primarily using it as a replacement for satellite backhaul on very remote cell sites. That way you don't need residential roof antennas and a ton of power to communicate with the network, just need your main power at the cell site. That's more practical for poor rural countries but isn't as big of a writeoff, doesn't look as sexy.


I could see Starlink-fed WISPs or cellular becoming something very real. Starlink CPE is very expensive and probably heavily subsidized. Plenty of villages where 0 can afford it, but collectively they could.

Starlink may as well take the money on the table if the constellation will fly overhead anyway.


This is the business model that Telesat trying to make work. While they are behind Starlink in terms of rollout, they are a credible manufacturer/provider for a service like this

https://www.canwisp.ca/resources/Conference%202020/Presentat...


I take all that to mean that Telesat hasn’t cracked the $$$ CPE nut like SpaceX has, or the SpaceX CPE subsidies are even bigger than I can imagine.


Small point, but the idea here is not using additional spectrum, the antenna design lends (in their analysis) up 29dB added link budget. For a more intuitive example, that's roughly the equivalent of pointing a satellite dish at your phone.


>This is why in urban areas it usually makes more sense to use cell network to save battery

Hmm. Rhen why does Apple recommend using WiFi for power saving?

“When you use your device to access data, a Wi‑Fi connection uses less power than a cellular network”

https://www.apple.com/batteries/maximizing-performance/


There's a lot of confounding factors, but in general: Cell network uses very little power when idle, but a lot when transmitting. Wifi on the other hand uses a medium amount of power when idle, and a medium amount when transmitting. (this is really generalized and there have been a lot of recent advancements to optimize power usage of both).

This is why general guidance for years was to use wifi when you are actually using your phone for browsing the internet, taking advantage of wifi's lower power efficiency when in use, but to turn off wifi when your phone is just in your pocket/idle to take advantage of cell network's better idle power efficiency.

However, the above is really only true for urban/suburban areas where you have good signal. If you are in an area with poor signal, your phone will use a lot of battery power via the cell network even when "idle" (because in reality even though your phone is "idle", the cell radio isn't actually idle, it's constantly struggling in the background to maintain the connection through the poor signal).


Just adding onto your comment, Apple probably recommends use of wifi because: 1. They collect AP's for their mapping/location crap. 2. The wifi chipsets they use are basically their own self-contained computers that can manage scanning at low-power without taxing the CPU. 3. At least for new HW, some of the 5G-NR stuff really isn't well optimized yet, not as bad as LTE originally was... but it's entirely possible the new modems for now are a little power hungry. Turning on wifi makes the cell chipset track less carriers, typically it'll just hug to one low-band carrier for status/push messages instead of as many as 5 or 6 carriers.

EDIT: Phones usually only transmit on one channel while receiving anywhere from 1-7 different channels at once, only the recent flagships transmit on 2 or 3 right now.


> Apple probably recommends use of wifi because: 1. They collect AP's for their mapping/location crap.

Nope. And the reason this is obvious is because turning off WiFi via the control center doesn’t actually turn off WiFi so collecting SSIDs wouldn’t be a problem either way.


This sounds like an appropriate test for equipment deployed into large areas of Africa, Central Asia and the Middle East where T-Mobile/Sprint don't operate.


I don't think the problem is the companies but the band itself, which follows physical laws. GP has experience with the limits of this band because it's used by Sprint/T-Mobile.


Sure, but perhaps it's an improvement on any existing coverage options or even a lack of any.


I just don't buy it.

One way to make a system like this work is heavily proxy and strip rich media out of the network (Globalstar sells a device called the SatFi that does this for their MEO constellation, Opera Mini used to do it for featurephones), so you can support quality of service in a remote population.

There's a financial incentive for Facebook to provide access to a specially crafted 'walled garden' that's designed for lower bandwidth scenarios. Facebook has tried this shit before in India without touting special network hardware and it was not well received: https://www.cnet.com/news/facebooks-free-basics-internet-ser...


Yup, this is exactly "Free Basics” reloaded. Fortunately their sat blew up in their face and then Jio erased any competitive advantage they might have had in India.


But given that way more Indians now have data, I'm not sure that it's much of a loss to Facebook, as they got the users.


They dont have buildings there? or no trees?


Are you excited for Starlink, or will line of sight be a problem for that?


So long as you don't live in a crater or have skyscrapers blocking your view line of sight should not be an issue for a majority of the population. Starlink wouldn't require a specific southern view of the sky like a GEO system does but it won't do well in 'urban canyons'. MEO consumer things you might be familiar with that experience reduced signal in the described conditions include GNSS (GPS) and SiriusXM.


Starlink is not meant for densly populated areas where 'urban canyons' as you called them would be found. The bandwidth of the simultaneously reachable satellites is basically constant everywhere, except at higher latitudes where satellites on different orbital planes get closer to each. So distributed users are preferred to lots of users in a single place, since the latter would have to share the bandwidth while the same number of distributed users would access via different satellites instead and would see an increased total bandwidth available.

Look at the graph on "population and satellite density per degree" in [0] to see why SpaceX has most likely chosen the inclination that they use.

[0] https://caseyhandmer.wordpress.com/2019/11/02/starlink-is-a-...


Sirius is now GEO, but has ground repeaters in problem areas.

GPS has the advantage of seeing multiple satellites and working with that by design since its one way. Despite that, Japan launched its own satellite to supplement GPS because it had so many issues. I think we’re spoiled by our phones that supplement GPS with cellular signals, wifi and inertial references.


The newest phones actually even do dual-frequency GPS using a newer signal (L5). It's accurate enough to do lane-keeping.

Existing systems also supplement GPS network data with russia's GLONASS, as well as sometimes WAAS if the satellite gets picked up (WAAS is GEO system to provide for additional accuracy, since it's further and requires a path facing south, it's not often picked up on phones)


>Inertial reference

Can we really do inertial reference with current phone hardware? I would imagine our usecases don't really even give inertial reference a nice playing field.


>Can we really do inertial reference with current phone hardware?

Yes, check the specs for any recent (past 3 years) qualcomm chipset, initia sensors will be included. You can play around with this sensor (and others) using the Phyphox app for ios and android.


Maybe my terminology isn’t ideal, but I figured my NAV apps can tell I’m doing things like a U-Turn based on G-forces. But maybe it’s just using its compass.

My old Garmin wouldn’t realize I did a u-turn until I have half way back down the street.


I'm excited about the quantum teleportation fermilab has been making progress with, but I don't know if line of sight will be a problem for that either.


CSIRO's Ngara aims to solve the same problem of rural access [1]. It was designed about 10 years ago, with a view to being the rural segment of Australia's National Broadband Network, but the NBN passed on the technology.

Ngara was designed to work in the vacated VHF/UHF TV bands (but will work in other bands). It uses MIMO with a large number (~32) of antennas to do spatial multiplexing rather than FB's "brute force" sectoring.

Sectoring is a disadvantage of FB's method, as the sectoring will limit the multipath and the ability to do spatial multiplexing via MIMO. With sectoring you will only get full capacity if your terminals are evenly distributed in azimuth. With MIMO you can get full capacity as long as the antenna has a big enough aperture to resolve each user, which will work better if you happen to have a small settlement in the cell. MIMO can also trade spatial multiplexing for diversity, which will allow longer range if there are fewer users.

FB could do a lot worse than just buying Ngara.

[1] https://research.csiro.au/wireless/888-2/


Gotta make sure rural folks have their Facebook news!


*ads.

Totally love it when advertising companies pretend that they're making the world better by trying to be ISPs where people don't have good networks. Google did it too iirc.


"It is not from the benevolence of the butcher, the brewer, or the baker that we expect our dinner, but from their regard to their own self-interest"

-Adam Smith


Grocers are not natural monopolies.

Adam Smith died before John Stuart Mill coined the concept, and a full century before electricity was invented.


Similarly, he also didn't blindly support the idea of free markets solving every problem, nor ignored that they have their limitations, in the way that he is portrayed by the people selectively quoting him.


The issue comes starts we start regulating and trusting baker's like brewers and butcher's like doctors. If I am in the market for beer then the brewer is fine but if I'm in the market for bread I sure hope it has not fermented.

Similarly, if I am in need of ad subsidized internet then fine but I would not want additional public funding or regulatory capture to assist Facebook.


> but if I'm in the market for bread I sure hope it has not fermented.

Because you insist on only eating unleavened breads?

https://vinepair.com/articles/want-to-understand-fermentatio...


If you substitute “bread” for “matzah” (unleavened bread) the whole comment could be a weird jiddish joke.


I read the parent comment, and thought that it was a matzo joke.


Facebook has trod this path before too, with their solar powered ultralight 'pseudo-satellite' drone. That predictably didn't work.


win win situation? With net neutrality, that is.


Ah yea, I remember all that drama and fear mongering that amounted to nothing .


Comcast will have nationwide data caps in two months...


This doesn’t compare to the most common approach to long haul wireless in rural areas.

There are many competitors that use an omnidirectional tower and directional antennas at customer sites.

A recent competitor in this space regularly achieves 6 miles at >100Mbps (~16mbps uplink):

https://www.telecompetitor.com/wisps-get-cbrs-range-as-great...

FB is claiming 6Mbps down (1.2 up) at 12 miles for their approach, which doesn’t seem to rely on fixed location customer antennas.


The antenna FB is describing provides cellphone coverage. Wireless providers just provide Internet/network capacity (since they rely on a customer's fixed antenna). Both those things are needed in rural areas.

Both these solutions require that you be in line-of-sight to the tower. You are more likely to be in line-of-sight to 250 meter tower.

So, you're right, not comparable but still.


I guess if you're a terrestial teleco you have to do something to compete with Starlink, but I'm not sure I'd make the investment in these in rural areas if SpaceX can really make Starlink economical. But that's a big IF right now.


Paradoxically, 'Starlink' may have more measurable economic value for unintended customers rather than 'rural folks' etc..

For example, urban homes and companies may want to chose them for any number of reasons and if they do, their willingness to pay may be higher than for rural folks and definitely higher than rural folks in Africa.

There probably is no market-based way to to do such a thing if there are global bidders for a resource unless there's some kind of non-market intervention such as Starlink setting aside a certain bandwidth per nation, or per economic tranche etc.

Novel solutions to the rural/distance problem are welcome. Even it's just literally 'a higher tower' as long as it works and is sustainable/cheap.


StarLink is taking FCC rural bandwidth funds, so there is a geographic tranche they’re prioritizing/reserving capacity for.


Starlink won't be able to provide service to all urban customers at the same scale as landline ISPs can. Here and there there they can take a customer but there wont be enough spectrum to serve everyone.

Also if you have to put up a pizza sized box outside your home, your landlord might complain. Even if you are the owner, building codes may prevent you.


Rural Africans won’t compete with Americans. The satellites over Africa aren’t serving Americans at the same time so they are different markets.


I’m ready to jump off of Comcast as soon as possible, but Starlink may not be the best option there. Now if I was working from an RV all the time it would make tons of sense.


Because traffic will downlink traffic locally (for multiple reasons) there should be virtually no interference between countries.


Does Starlink provide a mobile phone service?

BTW any internet provider could, they usually offer a SIP phone anyway. If this could be connected to a femto-cell coveting a household, this could solve the problem of using a phone around the house, in the backyard, etc.

Making it accept a SIM card from an existing provider is another problem.


I thought "wifi calling" solved that long ago. It's been enabled on my iPhone for a few years now and works with att.


Would be dead simple for them to offer a Google Voice like app that links to a standard terrestrial phone network backbone.


It's just a tall tower with more antennas. Those aren't rare. American Tower's tallest tower, in New York State, is 358m. There's no problem doing this. The question is whether to build fewer large towers or more smaller ones, which is an economic tradeoff.


Upvoted because you aren't exactly wrong. femto's comment explains the logic, they combined a bunch of splits in one antenna unit so they can direct spots. Femto also explains why there's a large disadvantage to doing this in the real world where things like towns/villages exist instead of perfectly spread out groups of people.


This is the sort of thing that moves the world forward, but Facebook don't seem to have built a tower to test this. They seem to have used pre-existing towers and modelling.

Telcos would have known this works but haven't done it because of why? Regulations on high towers? They don't want to be stuck managing these unique things? What's the catch.

The solution is probably not technology but some sort of process issue that you need to actually get dirty to find out about and fix. Facebook needs to invest in building towers. Go do it in Africa. If it's profitable they wont lose money and if not they've tried something great.


The ISPs and network manufacturers know all of this stuff. They likely already have superior modelling ready.

It is just not profitable. The towers cost money, the equipment inside those towers cost money, the power to towers cost money, if not using radio-links the network to towers cost money, the core network in these rural areas cost money...

And really you can get only so much money out from single tower. Considering how plan prices are fixed for cell services it could be that even over decades of time such installation won't pay itself back.

If the service providers thought they could make money by offering service to area they would be doing it.


I think people are beginning to see through these advertising companies. If it’s too good to be true it probably is...



What's the issue? If you're implying there is a trademark problem, it's not clear. Products/companies can use the same name.


exact same name? I thought you could use a similar name if you were not in the same market but I didn't know that you could use the exact same name... I guess Facebook has a bigger legal budget.


You can use the same name as long as it’s in clearly different markets. However, plenty of issues can crop up. https://en.wikipedia.org/wiki/Apple_Corps_v_Apple_Computer


You can use the exact same name and filing a trademark actually is more useful to help existing trademark holders delineate and remove ambiguity. It at least removes their claim as they obsessively try to defend how they understand the world.

Trademark filings let you mention that areas of trade and commerce you want the trademark for.


Could I trademark “Facebook”, a banana export company?


If others - such as Facebook Inc. - didn't already trademark it in agriculture and export, and you are prepared to defend it then sure.


See Lexis-Nexus vs Toyota Lexus.


Do you mean LexisNexis?


I can't help thinking that reliance on a smaller number of these SuperCell towers will make the network less distributed and therefore more vulnerable. For example, attacking a smaller cell may not be worth the effort for a malicious actor because impact of outage is very limited. It sounds like with SuperCell you can take down one tower and the impact is similar to taking down hundreds of sites.

My intuition may be wrong here as I'm not an expert in cell networks, but the tradeoff seems quite analogous to ones faced in distributed computing, storage and other systems. I'm curious to see some more authoritative opinions on this.


Can anyone comment on if there's anything novel here other than the config i.e. tower height + sectors?


They tried similar thing in India and it was shutdown because it violated net neutrality laws.


For the Luneburg lens antennas, are they using MatSing (https://matsing.com/) balls? They look a lot like them, especially the rear part where the RF cables terminate....


I've seen a number of these types of projects crop up in the northern Swedish forrests during the 2000s.

Every single one that I heard of petered out or was revealed to be a straight up scam. Like trying to sell wifi to rural homes in the woods.


Whatever happened to the Cantenna concept?

https://www.makeuseof.com/tag/how-to-make-a-wifi-antenna-out...


Cantennas were always a cheap and nasty kludge.

Their performance is poor compared with yagi's, dishes, etc.

Plus the name is doubly ironic. To an RF person, a Cantenna is a homemade dummy load which is specifically intended to not radiate.


Cantennas were always inferior to real yagi or dish antennas which are now <$100.


The Eye of Zuckerberg?


I read this as a tacit endorsement of socialized network infrastructure.


TLDR: Made a bigger tower, it worked better.


While serving more ads! I remember reading in Ready Player One about IOI controlling most people's internet connections and mentally drawing a parallel to Google at the time, thinking that would be a huge monopoly if Google not only served the online services, the laptop hardware, the laptop software, and now also the uplink itself. And indeed, a few months after I read the book, they announced the fiber project. Then Google and Facebook were going to give people in Africa internet with balloons and stuff. Now we can add this initiative.

It's really getting there: ISPs are shifting from a specialized business (they may do TV, web hosting, etc. as well, but it's always related services together) to something that comes bundled with your favorite ad service. They have the money to outcompete any existing ISP they like, so long as they somehow manage to dodge anti-anticompetition laws they can really give the non-ad-supported services a tough time. Ads is were we're headed, it would seem.


Peak capitalism: when ad companies own your communications infrastructure.


Now with the added bonus of taking your personal data, location, etc, because it's in your contract.


I remain baffled how many people I know who run ad blockers cheerfully bought Google routers. And FB is arguably worse than Google here; I can’t imagine willfully handing FB all my network traffic.


I understand how networking works and I think for myself so yeah, I have a Google router. I'll be chuckling quietly when your router gets wormed.


What assurance do you have by understanding how networking works? Have you actually dug into the software on that Google router to make sure it isn't backhauling the output of conntrack -L, perhaps for quality assurance and product feedback purposes?


What assurance do you have that any closed-source router isn't doing that? It's all the same. If anything, FB/Google have more to lose than your average company


None, which is why I don't use closed source routers.

Between different companies is a subjective opinion. I'd argue that people who trust FB/Google have already bought into surveillance culture, and so those companies don't have much to lose by adding network traffic to their take.

Reasonable people can agree to disagree about that, but it's just odd to claim that "understand[ing] how networking works" makes for some kind of objective judgement.


I've mentioned this on HN before, but I find condemnation of the NSA situation followed by calls for municipal broadband similarly baffling.


If we were talking about some federal, nationalized internet infrastructure, I'd agree. Municipal broadband isn't that. The NSA operates at a much different scale than a municipal broadband provider, which is small and basically transparent. Local broadband providers have no capability to undermine cryptographic standards, use libraries of 0days to take over your PC, or anything like that. NSA et al aren't going to grant them that capability either. I'd at least like the option of an ISP that is under indirect voter control, to complement the usual 900lb gorillas that line their pockets at everyone's expense and are basically unaccountable.


No municipality has the competence to tap your connection.


I'd be more pragmatically worried about XfiniWarner than a municipal broadband provider. The former operates at scale and can escape accountability by fallaciously invoking private property, whereas the latter has at least some democratic oversight at the town level. Although really all communications companies are effectively organs of the state, regardless of their governing charter.


Municipal agencies are subject to a lot of safeguards e.g. FOIA that private companies are not - market competition is supposed to hold them accountable instead, but for natural monopolies like internet service the competition doesn't actually exist.


First they say 5G isn’t dangerous, and now they say a 5G supercell is going to precipitate in our neighborhoods.


Like rain or like snow?




Consider applying for YC's Spring batch! Applications are open till Feb 11.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: