Facebook demonstrates record-breaking data rate using millimeter-wave technology | Hacker News Reader

Facebook demonstrates record-breaking data rate using millimeter-wave technology

code.facebook.com

111 pointsfforflo59 comments

The article mentions that water vapor (humidity, fog, clouds, etc.) is the most significant source of attenuation at these frequencies/wavelengths. This is also the most important source of attenuation for radio astronomy observations as well, along with atomic oxygen and ozone. This is why millimeter-wave observatories have to be located in high altitude (> 10,000 feet) and dry locations, like the practical moonscape of the Atacama Plateau in Chile where the Atacama Large Millimeter Array (ALMA) [1] is. ALMA is a truly amazing instrument that's transforming modern astronomy, and they operate between 4 mm to 0.350 mm (70 to 890 GHz).

If you want to see the effect of water vapor on these sorts of astronomical observations, take a look at a nearby telescopes transmission calculator [2], where it shows the percent of light that's transmitted from space to the ground as a function of frequency. The "PWV" means precipitable water vapor, and is the height of a column of water from precipitating all of the water in a 1 centimeter square extended from the ground all the way up. The Atacama is a super dry site, and 1 millimeter of PWV is routine there, but try putting in 100-1000 GHz with 0.5, 2, and 5 millimeters PWV that are typical of other observatories!

[1] http://www.almaobservatory.org/

[2] http://www.apex-telescope.org/sites/chajnantor/atmosphere/tr...

They conveniently tested it in SoCal.

> Specifically, rain and humidity could result in very high attenuation for terrestrial MMW links and could severely compromise their availability.

I imagine this will be useless in most equatorial parts of the world, and probably even south-east US.

It's unfortunate that they didn't delve deeper into the atmospheric attenuation since the whole point of the project is to deliver internet from UAVs. Seems like the next test should involve a foggy day in San Francisco instead of a "air-to-ground bidirectional link".

some tangentially related systems have fallbacks to different wavelengths that penetrate better in sub-optimal conditions. I don't know if it's applicable here, but I imagine it could be.

Baby steps. They weren't sure if they could even fly a solar plane, so they figured that out first. Then they weren't sure if they could even communicate with the thing once it's up in the air, so they've just figured that part out. Next step will be to figure out if they can fly and communicate in different weather conditions. And maybe they won't figure that out, but at least they didn't "imagine it was useless" and not even try.

I guess you missed the part where they say they set up a duplicate at a data Center that gets a lot of rain (Prineville, OR in this case)

I didn't read the statement where they discuss this, but Prineville, OR does not get a lot of rain. It is in the eastern Oregon high desert, very near Bend, and it gets more than 200 days per year that are sunny or mostly sunny, the highest number of clear sky days in the state.

In recent years they substantially fixed Ku-Band communications in poor weather with some new algorithms and consequent hardware/protocols. I suspect this would be the next step for this technology.

Not really. The fundamental problem is still there where the attenuation is high at certain frequencies in C/Ku/Ka. The way they mitigate it is to use different modulations and code rates in each frame so you don't need to bring the quality of the entire link down due to one user. However, the tradeoff is that the system capacity goes down fairly sharply as a result of this.

Sure, but there are plenty of parts of the world where this actually would be extremely useful. Complaining that a new technology isn't useful in 100% of cases is such an unhelpful position to take. Finding a solution that works in 10% of the world means the next solution only needs to worry about 90%. No one ever claimed that there would be one single solution to the problem they're attacking.

So wait... is the defense industry involved in the Aquila project somehow? The applications to defense/surveillance are obvious, and the aerospace technology (if not the electronics) has been developed for the military/CIA/whatever.

Is Facebook getting government money to develop this? Or hoping for that as a market? (Or is it, "We've already got one"?)

By accepting military money, you usually have to sign off that you won't use or sell the tech outside the USA. It usually comes under ITAR regulations too.

When you're Facebook, and your main target market for new connectivity technology is Brazil, Indonesia, India, Malaysia, etc, that tends to be problematic.

1mm wavelength is also known as 300 GHz carrier frequency.

The article states they use 30 - 300 GHz carrier with 2GHz bands. Definitely line of sight technology. The graphic shows airplane use with a 30-50 km range. Possibly usable in balloon relays too? (not satellite distances)

Links are in the 30-40 Gbps throughput range.

> To put this in perspective, our demonstrated capacity is enough data to stream almost 1,000 ultra-high-definition videos at the same time.

Is FB considering getting into the TV market? This would be a way to enter the market and compete with Dish, Time Warner, ATT, etc.

Facebook live seems to be doing pretty well so this seems likely.

Would this work for wireless vr?

That's exactly my thought. Short wave wireless to enable extremely high throughput things like uncompressed video (as if it's a wireless HDMI cable) as well as complex signals like accelerometer telemetry, etc mean this is a perfect technology to make the Oculus wireless.

Befuddling things: Trump defeating Clinton, Highly-disparaged MacBook Pro outselling highly-acclaimed Surface Book in five days, Facebook (who uses facebook anymore? who ever trusted them?) continually developing and deploying the most awesome commercial tech in the world.

EDIT (Yes, to the comments, duh lots of people use facebook, buy disparaged MBP the day it comes out and vote whatever. Some don't even use an oxford comma. I just don't get it.)

that's news to me that MBP is outselling Surface. Also, at least 1 billion people use Facebook so that must count for something.

It's not a secret that Surface has not been selling well. Actually maybe it is, because the numbers have to be derived by working backwards by dividing revenue with ASP. I think it was something like a million units in the last quarter.

1.7 billion people use Facebook.

That is extremely likely to be an inflated estimate.

Why?

Fake accounts among other things. For example ~0.1% of their users die every month but they often don't get notified etc.

Further, FB's own estimates cover a fairly wide range, but they have little incentive to user lower estimates.

By their official numbers, they have 1,18 Bi DAUs and 1,7 Bi MAUs.

If somebody dies, they won't be counted as a DAU anymore and in on month won't be counted as a MAU.

More numbers in their Q3 earnings reports: https://investor.fb.com/investor-news/press-release-details/...

This may seem pedantic. However, it's not uncommon for a relative to log into a dead users account. Thus showing up on DAU and MAU even after death. Users will log in if for no other reason than to notify people that the user passed and mention the funeral date etc. What makes this significant is these accounts where real and FB really can't get accurate numbers for this so it's all internal estimates.

So, yes FB's official numbers take this and many other factors into account. However, if their internal estimate is say 5%-40% of accounts are fake they are likely to use a number close to 5% than 40%.

While that may happen, the frequency is so low (and would happen once, not every month), that it does not materially impact numbers in the billions.

Dead users are just one of many types of 'fake' user. On it's own it's a small percentage of total active accounts.

But, I bring it up as the proxy for lot's of other similar estimates. I don't know if what % of 'active' accounts in their logs be that 0.1%, 0.2%, or 0.02% are dead, but I suspect they are going to pick a subjective number that makes them look better.

Further, because actuate numbers take a lot of accurate estimates even 'small' biasing of these sub estimates will compound and heavily bias their reported statistics.

AKA, spam accounts may be easy to find. But, spammers may create other fake accounts to mask their 'spam' accounts.

> who uses facebook anymore?

1.18 billion daily active users on average for September 2016

1.09 billion mobile daily active users on average for September 2016

1.79 billion monthly active users as of September 30, 2016

1.66 billion mobile monthly active users as of September 30, 2016

http://newsroom.fb.com/company-info/ :)

What is a daily active user?

If I visit a news site with a "share on facebook" widget which uses a facebook endpoint to get my facebook identity, am I an active facebook user?

If I have the mobile app installed which runs in the background getting notifications I never read, am I an active facebook user?

No. They are very conservative with their numbers. A Facebook user visits the website or the mobile app to count.

How do you know this? Can you back it up?

http://www.adweek.com/socialtimes/monthly-active-users-defin...

"We define a monthly active user as a registered Facebook user who logged in and visited Facebook through our website or a mobile device, or used our Messenger app (and is also a registered Facebook user), in the last 30 days as of the date of measurement."

That says nothing about FB erring on the side of conservativism.

From the the link:

'The social network said in an email to SocialTimes that it “wanted to provide a cleaner metric,” so starting with its third-quarter-2015 earnings report, it no longer counts what it called “third-party pings,” or people who did not access Facebook itself, but shared content or activity via Facebook Login-integrated websites or applications.'

That's exactly what is under discussion: whether FB conservatively excludes 3rd-party widgets and such when counting activity. They started being more conservative late last year.

It's almost like you're taking your personal experiences and extrapolating them to the entire world. Step out of your bubble.

I'm completely with you, but how?

Follow people on Twitter who you'd never meet in real life, people from different backgrounds or with wildly different points of view.

There's no bubble on Twitter but the one you make for yourself.

Unfortunately, stepping out of my bubble on Twitter would involved reading racist, sexist, homophobic, Islamophobic, xenophobic invectives. I know we're entering the "post-PC" / "PC-backlash" era but that doesn't mean I ought to spend my precious attention on people whose opinions are not worth considering.

But, then again, I've never really been on the Twitter train anyway.

Either you want out of the bubble or you don't.

There are people with contrarian views that aren't necessarily completely batshit insane, like those that chug Alex Jones.

For example, I've found David Brin to be fairly balanced, far from left-leaning, but not the sort of Libertarian people like Theil aspire to be: https://twitter.com/DavidBrin

I'm sure other people can post recommendations here.

> Either you want out of the bubble or you don't.

I don't think that's a fair assertion. It's possible to both want to understand opposing points of view but also not have the bandwidth to filter out clearly not-worthwhile text.

I'm never unblocking comments on YouTube. I'm never going back to Reddit.

It's like being home for Thanksgiving: When the conversation starts with "Well Obama's a Muslim so..." you know it's time to take your turkey plate to the living room and watch the parade.

I'm curious how you think one can understand without spending bandwidth.

As painful as it is you have to engage with people like that. If your own family is chugging Trump, it's a chance to try and nudge them, incrementally, away from their extremist positions.

Like maybe admit that Hillary might not be literally the devil. Baby steps.

I can't tell with this particular topic - are you making a bad joke or do you legitimately feel this sort of smug behavior would be helpful?

I'm saying if you've got a family so far gone they're doing lines of Alex Jones on their Facebook page every day then there's ways of getting them rehabilitated.

This isn't smug. This is fact. Facebook is filled with toxic garbage. Simply nudging people away from that and to something more rational would be a huge step towards civil discourse.

You can't. But not everyone deserves my bandwidth. Empathy and intellectual curiosity are my only prerequisites. If a person can't meet those, I don't have the headspace for anything they have to say.

The less energy I spend being my own gatekeeper, the more I have for things that truly enrich my life.

I completely agree on twitter. It's facebook, mbp, and trump that I don't get. Amazingly, in roughly that order.

Taking a trip to a tourist destination such as Memphis or Nashville would be a good opportunity. Spending a day in the Central Valley or even Gilroy would be a start; interacting with people in real life helps you to understand their point-of-view better.

I vehemently disagree with the suggestion of following accounts outside your bubble on Twitter; most people are snarky and mean. It would most likely alienate you from them and never actually discover their point of view.

Btw, I dig your humility.

Haha, I lived in CA and it was too liberal (note my aversion to all-thins-Apple?). I lived in Houston for seven years and, well.... I've also lived in India, Korea, Germany, Scotland, Taiwan, Malaysia FWIW. I think I can deal with twitter. My soul is somewhere between MI and OR.

I'm guessing you spent no time in the Central Valley or north of Sacramento. It is very conservative in contrast to the Bay Area.

Colorado strikes a good balance. Appropriate levels of social spending, focus on fitness, very little finger-wagging by social conservatives, excellent opportunities for recreational shooting.

Welcome to reality

If this were sponsored and implemented by a trading firm there would be angry comments about leeches living off honest people's incomes. minimising latency to push another ladBible video into your eye balls is way more moral and better for society apparently.

Yeah, I'm going to believe that a company that basically lives on doing A/B testing on engagement in a data sharing GUI (facebook.com) with a hobby in PR-creating "hardware innovation" has suddenly beaten about a hundred thousand world-class RF/EE engineers in e.g. Ericsson, Nokia-Siemens, Qualcomm, Huawei, Alcatel-Lucent and Samsung.

The only way FB beats these guys is in marketing reach.

Since those other companies don't talk much about what's in their labs, for all we know they are already ahead of Facebook. Also, consider that Facebook may have hired people with years or decades of experience from those very companies.

Well, this is going to get them to talk about their progress in this field, at least. :)

Maybe. Or maybe the legacy vendors will double down on their old "you need standards compliance, backwards compatibility, support contracts, and a sales channel" line.

I think you're mixing things up.

a) These companies do love to brag about technical accomplishments. They each have hundreds of product marketing and engineering people for this purpose alone.

b) When they do sell, they will indeed double down on the things you mentioned.

So you're unaware of the open data center work that they have open sourced, which is innovative in countless ways?

You're speaking of their open source mechanical design of a rack-mounted PC? Yeah, relatively speaking that is entirely unsubstantial.

The mere fact that you're comparing this to the amount of research that goes into radio/network protocol design by the big companies is hilarious, I have to say. But also cute. Do stay ambitious :).

Reply on news.ycombinator.com