Alphabet's Project Taara beamed 700TB of fiber-like Internet with 99.9 percent uptime

nanoguy

Posts: 1,355   +27
Staff member
Why it matters: For years, companies like Google and Facebook have been exploring ways to bring Internet connectivity solutions to people in remote areas around the world. The latest attempt uses 20 Gbps wireless optical links that were developed for Google's Project Loon to connect communities where fiber infrastructure is too complex and expensive to implement.

Google has explored plenty of daring projects over the years, but most of them never see any commercial implementation beyond a few trial runs. That's because the company has cultivated a culture of embracing failure as a way to test ideas and quickly kill the terrible ones before they suck in too many resources.

A notable example comes to mind -- Project Loon, which was dissolved in January after Google found that sending helium balloons into the stratosphere to distribute wireless internet to more than a billion people living in remote areas wasn't commercially viable.

However, some underlying technology has been preserved for use in other projects at Google's X labs. Specifically, the company has been using Free Space Optical Communications (FSOC) links for a little-known experiment called Project Taara.

The idea behind Taara is that expensive fiber optic installations can be replaced with 20 Gbps FSOC links that beam connectivity over short distances. The project has already seen a few pilot implementations in Kenya and India, and today the company revealed that it's also been using this wireless optical link to connect two communities on the sides of the Congo River -- Brazzaville in the Republic of Congo and Kinshasa in the Democratic Republic of Congo, totaling over 17 million people.

Baris Erkmen, the Project Taara lead, says the link successfully transmitted around 700 terabytes of data in 20 days of operation with a respectable 99.9 percent uptime. Making this optical tech work over a distance of 3 miles (4.8 kilometers) is no small task, as it requires constant minute adjustments to align the laser beam with the receiving node.

To that end, the nodes are placed high up and can automatically adjust their mirror system to maintain a steady connection even with optical interference from birds and changes in weather conditions. The Project Taara team has spent countless hours improving the reliability of FSOC nodes to the point where the engineers are confident they can operate with minimal service interruptions.

As you'd expect, this system wouldn't work in areas where there's constant fog, such as the San Francisco Bay area. That's why the Project Taara team chose to conduct these tests in a place where the climate allows for a relatively clear line of sight between the optical terminals.

However, the cost of this system is five times lower than trying to run a fiber optic cable 250 miles (400 kilometers) around the river, or laying the infrastructure needed for 5G connectivity. And while it won't be suitable for every location, it does have the potential to connect hundreds of millions of people in remote areas to the internet in a more cost-effective way. Red areas in the above map indicate places where Google thinks the technology can be implemented for over 99 percent uptime.

Overall, the lesson learned by Taara's engineers is that you can sacrifice a small amount of signal reliability to bring fiber-like Internet speeds to communities of people that would otherwise not be able to afford the cost of a wired backbone. To them, this new technology would be indistinguishable from the traditional broadband that most of us enjoy in sprawling urban centers.

The Project Taara team says it's currently exploring partnerships with governments and telecom operators to accelerate development and implement the wireless optical link in more areas around the world.

Permalink to story.

 
Alphabet, Google's parent company, should not be trusted and allowed to control internet. They ARE worse than Huawei (but no one dares talk about it because they might commit suicide)
 
I wonder how much (if any) improvement would be achieved if they doubled or tripled the size of the receiver target? There has to be a point in size when the failure rate drops to an insignificant amount and still meets the cost vs. benefit goals .....
 
Pedantic question. Is five time cheaper easier to understand that saying 1/5 of the price?
People in general understand and parse multiplication faster and/or better than fractions (division). See: A&W 1/3lb burger. "Five times cheaper" arguably reads more cleanly than "one fifth the cost" as well, depending on the person.

I've had situations at work where giving a pitch to do something that costs time/money, an explanation by multiples elicited a more favorable response than explanation by parts. Do there's that anecdotal experience too.
 
Pedantic question. Is five time cheaper easier to understand that saying 1/5 of the price?
People in general understand and parse multiplication faster and/or better than fractions (division). See: A&W 1/3lb burger. "Five times cheaper" arguably reads more cleanly than "one fifth the cost" as well, depending on the person.

I've had situations at work where giving a pitch to do something that costs time/money, an explanation by multiples elicited a more favorable response than explanation by parts. Do there's that anecdotal experience too.
Anyone who has watched commercials in the past few years will also know that people like "x times" better than any other way to compare...

Instead of "twice as", you always hear "two times".

I see this simply as catering to the lowest common denominator - aka REALLY DUMB PEOPLE CAN'T DO SIMPLE MATH!
 
From this article, it sounds like Google dropped its efforts at their fiber initiative because it was too expensive to lay the fiber backbone. To me, this is really interesting because I recently got FTTH installed from a small local company laying the fiber backbone in our area. That company did, however, get an investment from a billionaire. Even with all their money, the great Google was unable to accomplish a similar feat? 🤷‍♂️
Google found that sending helium balloons into the stratosphere to distribute wireless internet to more than a billion people living in remote areas wasn't commercially viable.
Imagine that. 🤣
To that end, the nodes are placed high up and can automatically adjust their mirror system to maintain a steady connection even with optical interference from birds and changes in weather conditions. The Project Taara team has spent countless hours improving the reliability of FSOC nodes to the point where the engineers are confident they can operate with minimal service interruptions.
I wonder if that saying, "Always sure, often wrong" will come into play here. This seems to be a critical failure point in the system. I have to wonder when birds developed the ability to bend light? :confused: From the article, it is obvious that this system cannot handle things like fog that decrease signal. Will this be another project that Google dumps because of "limited economic viability?"

Personally, I think Google often outsmarts itself. More power to them if they actually make these installations work and make even more money that they don't need.
 
Last edited:
700 TB / 20 Days
= 35 TB / Day
= 1.45 TB / 1 h
= 1493.3 GB / 1 h
= 24.8 GB / 1 min
= 0.414GB / 1s
= 424 MB / 1s
Well, the article does say "Fiber like internet speeds". Interestingly enough, that per-second rate is lower than the lowest data rate that my FTTH provider offers which is 500MB/s, symmetric, for $50/mo.
 
That's how I interpreted it. That is the total throughput, so it's shared / split with everyone using it. So more like a 2400 baud modem :)
I would hate to share a 2400 baud modem in this day and age. :eek:

Something seems off even at that large of a data rate since the theoretical data carrying limit of light is on the order of 1/2 the frequency of the light. 424MB/s is rather low even for infrared light. It must be the electronics for the transmitter and the receiver - those electronics are likely not capable of driving the "pipe" to its fullest potential.
 
424MB/s >> 500mbps which is what your ISP is most likely providing.
Well, the article does say "Fiber like internet speeds". Interestingly enough, that per-second rate is lower than the lowest data rate that my FTTH provider offers which is 500MB/s, symmetric, for $50/mo.

Am I missing something here or are you both confusing "Megabits per second" and "Megabytes per second"

424MB/s (424 megabytes per second) is approximately 3.4 Gbps(3.4 Gigabits per second)

In the article they mentioned that this technology can reach speeds of 20 Gbps which would allow to download a file at a speed of 2.5 GB/s.

This is not bad for remote areas with not many people.
 
Back