Alphabet's Loon Balloons Just Beamed the Internet Across 620 Miles (zdnet.com)

Loon, the former Google X project and now independent Alphabet company, has developed an antenna system that could create a far greater ground coverage than previously possible. From a report: According to Loon each of its balloons, from 20km (12.4 miles) above earth, can cover an area of about 80km (49.7 miles) in diameter and serve about 1,000 users on the ground using an LTE connection. However, Loon balloons need a backhaul connection from an access point on the ground and without that connection the balloons can't provide connectivity to users on the ground. But on Tuesday the company revealed it had sent data across a network of seven balloons from a single ground connection spanning a distance of 1,000 kilometers, or about 621 miles. It also achieved its longest ever point-to-point link, sending data between two balloons over a distance of 600km (373 miles). The tests were carried out across California and Nevada, with the balloons punting data packets between each other from "desert to mountains and back again", according to Loon.

Comments

But how reliable is it?

[Anonymous Coward]

I am just looking at this from a standpoint of a lot of complexity that must all work for this connection to work.

Re:But how reliable is it?

[olsmeister]

The AI will take care of the complicated bits.

Re:But how reliable is it?

[shaitand]

I'm less concerned about that, the mesh network concept wouldn't really need to be that much more complex than many other similar networks that have been built and are well proven including those among satellites and smart devices you might have in your house now. Even removing points of failure and having the network be self healing should be quite manageable (note: that is without regard to actual balloons they are building as a platform which might be complex and unmanageable).

My concern would be scaling and future proofing. First there is the device and traffic chatter. The problem with most projects these days is that there are dozen scripting languages, libraries, automation platforms, development tools, etc. Each one takes months to learn to wield properly and far too many projects simply crank out with a complex blend of solutions from whichever ones devs on the team happen to have in their toolset. There simply isn't going to be anyone who can support it properly a few years down the road that way.

Then there is the issue that because complex frameworks and abstraction are all the rage the underlying mechanisms at the core of many of these systems of parallel operation and coordination have deadlocks, traffic light problems, etc that are masked or have bandaids over them and the people using the frameworks have embraced a philosophy of just trusting them because they are widely used... by a bunch of others making the same assumptions. We didn't have all these magical massive parallel and networked application frameworks in the past because we believed in understanding the implementations we built on and knew these flaws made such structures unreliable not because nobody could do it. Cascading failures and bitrot are inevitable with these kind of underlying flaws at large scale, they will appear at the edges as odd one off cases that go away with resets or the like but at scale in critical deployments it's statistically just a question of when the starts align and the house of cards collapses not if. I predict we'll see an increasing rate of critical enterprise failures over the next decade. Hopefully this isn't another one that will go on the list.

Re:But how reliable is it?

[Anonymous Coward]

Iridium and Iridium II are both more complex than this.

My question is 'Why aren't airline carriers already owning this market?

Re:But how reliable is it?

[HappyPsycho]

Plans are already in place working toward that goal https://newatlas.com/airborne-...

Re:But how reliable is it?

[HappyPsycho]

Such a network with such fast moving, constantly changing nodes is probably the worst case for building such a mesh, and as such is at least a couple years if not a decade or so away (I would love to be proven wrong). Projects like the balloons allow the engineers to isolate certain issues and work on others (In this case they seem to be focusing on the RF links between the balloons, the downlinks are most likely just re-purposed satellite earth stations) until eventually they can make the plane based network work at decent scale. For the planes one of the major issues is there are almost no constant "anchors" that can be used to bootstrap a client node entering the network, the distances we are talking about mean any sort of 802.11/DHCP style broadcast can't work without requiring allot of RF power on the client side. Maybe some sort of hybrid could work, with the balloons providing lower capacity links primarily for co-ordination and the bulk of the network capacity being provided by the planes?

I wouldn't fight too much with scaling it to ludicrous speeds vs going for complete coverage as realistically it's goal isn't to provide constant internet to a city but to rural / maybe suburban areas where normal high capacity broadband doesn't make monetary sense. One of the major limitations of satellite is not only the atmosphere but how expensive it is to refuel them, the satellite itself has to generate the downlink signal (think from a cell tower to your cell phone) and when its behind the earth has to do so completely running off of its internal batteries which need to last 10+ years to make the satellite profitable so transmit powers are limited hence most satellite dishes are so large (a huge monetary deterrent if required on the remote side of the link, most earth station dishes are naturally large to make up for signal losses in other parts of the system and the cost of that one dish spread out over the other links) and need to be aligned with allot of precision (even geostationary satellites are not very stationary).

So what?

[Anonymous Coward]

Satellite has got that beat in one hop and is probably faster even though satellite is terrible.

Re:So what?

[omnichad]

Nobody wants more aerial distance. That just increases latency. What matters is the useful transit distance from the ground station. The balloons are 12 miles high vs. about 22,000 miles high for geosynchronous satellites. Balloons decrease latency by several orders of magnitude.

Re:So what?

[110010001000]

Good point. I suggest putting the transmitters on top long poles in the ground. That would reduce latency quite a bit. You could tie the balloon to it too if you wanted to keep the balloon.

Re:So what?

[omnichad]

That would further reduce latency, but at the expense of having actual trees, mountains, etc. being in the way.

Re:So what?

[110010001000]

Genius!

Re:So what?

[Anonymous Coward]

That would further reduce latency, but at the expense of having actual trees, mountains, etc. being in the way.

One could always cut down all the trees and flatten the mountains.- free fire wood and more arable land. It's win-win for everyone.

Re:So what?

[Anonymous Coward]

Get rid of the people and you have no need for firewood or land, by your moron's logic.

Re:So what?

[Kjella]

Nobody wants more aerial distance. That just increases latency. What matters is the useful transit distance from the ground station. The balloons are 12 miles high vs. about 22,000 miles high for geosynchronous satellites. Balloons decrease latency by several orders of magnitude.

Well a bit more aerial distance wouldn't hurt. One ms at lightspeed = 186 miles so the plans Starlink has for 750 mile orbits is like 16 ms round trip instead of 500 to GEO and would be single-hop distances you'd need many balloons for that usually add about 4 ms/hop so in total I'm guessing as fast or faster. And being satellites they won't be affected by disasters on the ground. Of course at the moment it's just another over-optimistic Musk timeline but if we do get a LEO broadband network in place I think project Loon is dead.

Re:So what?

[retchdog]

1. it was a joke.
2. we're working on your suggestion anyway.

Re:So what?

[omnichad]

More hops isn't all bad. The last-mile downlink can only handle so many subscribers, while the uplink hop-to-hop backbone is one large interleaved connection. I am pretty sure that part of the reason satellite is so slow is not just the roundtrip latency but also capacity. The other thing is that the balloon network can probably mesh to some extent since there are more and selectively route traffic to less-busy paths.

I'm not sure what ground disasters are affecting you 12 miles up. Tops of hurricanes would get close, I suppose. Wildfire smoke doesn't have to go higher to interfere with either signal.

conspire to occupy the truth about us..

[Anonymous Coward]

cease fire stand down.. we rest our lower case.. thanks again..

AC fail

[bondsbw]

No claim was made that the area itself is 80 km.

According to Loon each of its balloons, from 20km (12.4 miles) above earth, can cover an area of about 80km (49.7 miles) in diameter

That is more meaningful than saying "5026.5 km^2".

Re:AC fail

[Anonymous Coward]

>quote>That is more meaningful than saying "5026.5 km^2".

The lack of understanding of significant digits is driving me nuts. You turned one into five! At least the article stopped at three.

This is more meaningful:
80 km diameter = 50 mile diameter = 5000 square km.

Re:AC fail

[EETech1]

12 meeelion acres...

Re:Measurements fail

[aevan]

Luckily, the editor foresaw your comment, and had used it as a measurement of the length of the diameter...knowing some overeager pendant would leap on it. Good bait.

Your school did teach you about the squareness of pies and finding the area, no?

Google will abuse their station at every chance

[Anonymous Coward]

They will use this to map unaffiliated networks and devices and track people outside of and alongside GPS, watch. It's a surveillance operation masked as a connectivity investment opportunity.

Data throughput? Latency?

[Anonymous Coward]

Marconi sent data across the Atlantic in 1903.

What is the throughput of the 620 km link? What is the latency? Sending data that far is not a huge accomplishment on its own.

One user for every two square miles.

[Anonymous Coward]

Fantastic for places where no people live.

Re:One user for every two square miles.

[Immerman]

That is indeed the initial target demographic from everything I've heard - heavily rural areas such as much of Africa, Asia, and the Americas, where many millions of people live at population densities too low to justify paved roads, much less short-range internet infrastructure.

Re:One user for every two square miles.

[retchdog]

wealth densities. population is just a poor proxy. if the residents mattered enough, they'd have those amenities.

Re:One user for every two square miles.

[Immerman]

Yes, because everyone is 100% convinced that crony capitalism accurately evaluates and rewards everyone's inherent worth.

And nobody would be interested in subsidizing an much cheaper way to deliver something as fundamentally empowering as internet access to places where most people have only heard of libraries in legend. Not out of egalitarianism, nor out of a desire to increase the taxable income of their population.

Re:One user for every two square miles.

[retchdog]

well, yes. but it's cheaper to just let them rot.

oh, and "crony" is superfluous here. you can just call it "capitalism".

Scaleable by orders of magnitude?

[Anonymous Coward]

This is a paltry data bucket for such a large area. Hopefully, it's just a proof-of-concept. I would really like another ISP in my area. Aside from increased data rate, what about hurricanes?

HSMM-Mesh has been doing this for years

[bobbied]

I'm sorry, this isn't some huge accomplishment.

Adhoc networking using HSMM-Mesh has been a reality on WRT54 hardware for YEARS. It can service multiple connection nodes, more than 1,000, including internet access if available to one or more nodes.

The flying of a GSM MSC/cell tower may be a bit less complex for the end user than having to have an HSMM-Mesh node to attach a network cable to but ham radio guys have been doing this on 2.4 GHz for years.

Also, flying a MSC/Cell Tower isn't all that unusual or novel either. We've been flying such things on fixed wing aircraft or in the back of trucks with crank up towers to provide emergency communications using cell phones for a long time too. Plus, flying HSMM-Mesh nodes on balloons has been done a lot too, to provide data network access to the balloon's GPS and cameras and run QSO's via data the data links.

Sorry Google, I'm not all that impressed..

Re:HSMM-Mesh has been doing this for years

[Anonymous Coward]

True. But i bet the haven't over spent on their system like google does....also they didn't mention and real data. like at what data rates? bps? kbps? Mbps?
So they provide a fancy LTE phone with almost GPRS capability?

Re:HSMM-Mesh has been doing this for years

[Anonymous Coward]

I think you failed to read the title of the article. It has nothing to do cell/radio technology. Google packaged up the physical, capital 'I', Internet and transported it, star trek style, 620 miles. This is truly amazing.

Seriously though, who writes this garbage and how come nobody holds them accountable for the destruction of our language?

Re:HSMM-Mesh has been doing this for years

[Bourdain]

Yes, the 620 miles sounds like the new/impressive part

Re:HSMM-Mesh has been doing this for years

[Anonymous Coward]

aiming a cantanna is hard?

Re:HSMM-Mesh has been doing this for years

[bobbied]

HSMM-Mesh works line of sight too. With 2W amps readily available, 600 Miles isn't some huge problem here. You may have to dial up the right parameters on the 802.11 link to account for the delay, but that's not hard with the HSSM-Mesh software.

Re:HSMM-Mesh has been doing this for years

[Anonymous Coward]

aiming a cantanna is hard?

From a balloon? Yes, obviously.

Re:HSMM-Mesh has been doing this for years

[bobbied]

HSMM-Mesh will do 802.11b/g speeds just fine. Which is pretty good for remote data links build with nodes which can be had for less than $50 each...

Re:HSMM-Mesh has been doing this for years

[Pascoea]

Say HSMM-Mesh again...

Re: HSMM-Mesh has been doing this for years

[Anonymous Coward]

I'm HSMM-Mesh, I'm HSMM-Mesh, I know I am, I'm sure I am, I'm HSMM-Mesh

Re:HSMM-Mesh has been doing this for years

[Anonymous Coward]

I'm sorry, this isn't some huge accomplishment.

May be not but watching telecom execs loose their shit just might be.

Re: HSMM-Mesh has been doing this for years

[Anonymous Coward]

How does it fair in noisy conditions? I'm looking at deploying a solar powered community lan in an urban area.

bobbied has been doing THIS whiny act for years

[Anonymous Coward]

"I'm not all that impressed.." - Google fails to impress unrealistic snowflake misconstruing the point of an article and making it into an demonstration of basic networking buzzwords for no purpose. News at 11?

Interesting idea

[Locke2005]

A balloon-based mesh network would require massive redundancy to be reliable... and wouldn't the nodes migrate over time? If the prevailing winds blow one direction, do you keep bringing them down on the coast and re-releasing them upwind? How much control over the altitude of the balloons do they have to try to maintain position?

Re:Interesting idea

[EETech1]

They've trained an AI to navigate the balloons by making them follow the weather patterns to stay in position.

It's actually quite impressive!

https://plus.google.com/+Proje...

https://www.engadget.com/2016/...

https://tech.slashdot.org/stor...

Re:Interesting idea

[EETech1]

Oops.. missing link

https://spectrum.ieee.org/tech...

The beacons of Gondor!

[mysidia]

sent data across a network of seven balloons from a single ground connection spanning a distance of 1,000 kilometers, or about 621 miles.

LoTR used this method first.

Doesn't sound very practical

[Solandri]

each of its balloons, from 20km (12.4 miles) above earth, can cover an area of about 80km (49.7 miles) in diameter and serve about 1,000 users on the ground using an LTE connection

80 km diameter = 40 km radius
A = pi*r^2
A = pi * (40 km)^2
A = 5024 km^2

That works out to only one user per every 5 square km. This is the reason cell towers are typically spaced 3-7 km apart in urban and suburban areas. You need them that close to support the typical density of users in a cell. Their actual range if you don't have many users is much larger. GSM is limited to 35 km because it uses timeslices - beyond that a phone's transmission would arrive in the next phone's timeslice. CDMA will work as far out as the phone and tower are able to "hear" each other, which is more likely to be limited by line of sight than by distance (a 30 m tower only gets you about 25 km range before it's blocked by the horizon).

It might be useful in developing countries which don't have many cellular users, but from what I understand even third world countries are rapidly deploying standard cellular networks since it's so much cheaper than stringing up wires. That leaves the only practical use in emergencies if you block regular people from being able to use it, only allowing emergency personnel's devices to connect.

This could change in the future as MIMO becomes more commonplace (it's included in the 5G standard. MIMO basically makes the signals and receivers directional, allowing multiple devices to use the same bandwidth without interfering with each other (too much).

Re:Doesn't sound very practical

[Anonymous Coward]

In places with more people, we already have cell towers. The balloons are for places where cell towers are not useful because there aren't enough people. Last week I drove from Nevada back to Seattle along a path that crossed some serious back country in NE California and eastern Oregon. There were multi-hour stretches with no cellphone signal whatsoever, let alone data connection, and little to no other traffic. Many places you don't even hear anything on the radio. It would be great to have a cell tower in the sky every 50 miles so you could get help if you break down or some such, but also for entertainment. There'd be fewer than 1,000 people in that circle but I'm sure some of them want to get on facebook.

Why is this news?

[Anonymous Coward]

This is nothing more than Google exercising marketing muscle to promote an investment they made for luddite press. Like Verizon touting mobile video calling with grandma. Big whoop tie doo.

Re:Fake JEWS loon's balloons

[corrosive_nf]

TL,DR

Yep - over black rock - burning man -

[Anonymous Coward]

yep, saw that balloon at altitude on flightradar24.. pretty cool.. launch out over the ocean and came inland over N. Nevada.. cool.

Re: Know your enemy the JEW

[Anonymous Coward]

Dats eggzachtly wut duh joooz wont u2tink

Re:Anyone else read that as

[Anonymous Coward]

Get my family friendly Goat C shirt! ~ CaptainDork