This would be a game changer. I live in a connectivity desert (Bay Area, California). Outside of downtown SF there are virtually no residential options for Fiber. My house is 100 yards from a fiber pump - the quote to run the line was $50,000 (the ISP would waive the first $30,000). When I complained to the tech, he said they only have 7 residential fiber customers in the bay area. If Facebook really can bring high speed Internet to California they could bring it anywhere!
Huge swathes of San Francisco have access to residential Fiber. Sonic started building it out several years ago in the western half of the city, beginning in the Sunset District, which forced AT&T to abandon their failed FTTN strategy and start offering direct fiber service as well.
It's slowly coming. I've enjoyed Sonic fiber in the residential Richmond District for 3 or so years already. Sonic is still building out fiber in other neighborhoods (e.g. Marina and Cow Hollow are now coming online, I think), as is AT&T, but there are still considerable gaps where utility poles are oversubscribed. A single unusable pole can prevent reaching several city blocks simply because the replacement cost is so high it makes those remaining blocks unviable economically. In other neighborhoods with buried utilities, the city refuses to provide any assistance: no micro-trenching permitted, and no leaning on existing license holders to lease any pre-existing underground infrastructure.
Even the least dense areas of San Francisco are more dense than elsewhere in the Bay Area. Sonic provides residential fiber in many Bay Area cities, but rollouts are far slower. Regulatory costs (including review time) and less favorable RoI (less density) means it's taking much longer to build, even when theoretically profitable. Sonic doesn't have as much capital, and those with capital (e.g. AT&T) aren't very motivated to build new infrastructure, especially when they're already the incumbent. About two years ago I chatted with an AT&T linesman putting up fiber on the utility poles outside my house and he flat-out said he wouldn't be there if Sonic hadn't forced AT&T's hand.
EDIT: Sonic is further along than I thought: "Santa Rosa-based provider Sonic has deployed fiber to more than one-third of homes in San Francisco over the past five years, Sonic CEO and co-founder Dane Jasper tells SF Weekly." https://www.sfweekly.com/news/can-san-francisco-finally-clos...
I have a friend in SF who has been given the runaround by Sonic. He believes that Sonic's ticketing system is somehow integrated with AT&T and that AT&T is closing his install ticket after they finish an install at a different residence.
The City of SF bid on the utility infrastructure in the city that was formerly owned by PG&E. Hopefully this will allow the city to streamline installation of pole-based fiber installs, but I'm not sure how much PG&E ownership was causing problems before the auction.
Sonic has resold AT&T DSL and, more recently, FTTN services for many years. I think they also recently began reselling AT&T's fiber service in places where Sonic can't build out their own network for technical or economic reasons. See https://forums.sonic.net/viewtopic.php?f=10&t=6798 (User dane is the Sonic CEO and he's very active in the forums.) It's not Sonic Fiber if it's not $49.99 for 1Gbps. Their resold AT&T service has pricing closer to AT&Ts.
If that 1/3 figure includes resold AT&T service, that would definitely be less impressive.
I'm not too optimistic about any city-planned communications infrastructure. The "progressives" hold onto a pipe dream of a free, city-wide public network[1], and they seem to balk at partnering with private companies for this purpose. Fortunately the "progressives" usually fall [just] short of majority control of the city council, and haven't won the mayorship in over 20 years, but they still manage to throw monkey wrenches into sensible plans.
[1] There was (still is?) a master plan to bury fiber with the city's water & sewer network, but I think the city is half done with a city-wide, multi-billion dollar water & sewer refurbishment and replacement project without having buried any fiber at all. I'm sure most people in the city, including myself, would love to see a public fiber built-out, but it's just not financially viable. The city has too much on its plate, and the situation won't ever change. The self-styled "progressive" faction is all talk, and stonewall everything else with feigned outrage, much like national Republicans.
I wonder if a good rule would be if a new entrant offers FTTP to an area, an incumbent cant do the same for 15 years.
This way there is a huge incentive for startups as the incumbent cant just follow them and do them out of business while ignoring other areas. At the same time it puts pressure on the imcumbant to roll out FTTP to reduce startups doing roll-out.
Probably also consider something similar with pricing.
I'm in San Jose and am very fortunate to have AT&T Fiber aerials constructed in my backyard a few months after I moved here from Santa Clara. (I feel your pain.) AT&T now has quite a few customers in my neighborhood because they are faster and cheaper than Xfinity.
Unfortunately, I believe this install was one of the last under the DIRECTV merger deal which required AT&T to deploy residential fiber.
The spools and amount of manual work to install the fiber lines was really something to see over the course of 3 months and I can really see the value of an automated robotic fiber deployment. The legacy cables are really huge.
I don't entirely grok how the Bay can be the IT capital of the world and also have such poor fiber penetration. I've experienced commercial fiber first-hand in downtown SF, but that was never an option for residential use. Only Comcast, Spectrum and a couple of smaller mom-and-pop ISPs mostly focusing on WLAN.
I recently got a quote to hook up to CenturyLink business fiber for a church in a suburb of Minneapolis (on an industrial road). I believe the distance was around 700 feet to their termination point. With a 5 year contract, it was going to be at least $700/month. They told me the cost was due to getting all the permits needed and then doing the actual run of fiber (digging). Oh, and it would take 3 months to get done. They also told me I could bring that cost down if I recruited other businesses between the church and the termination point.
Yeah - that’s actually not a bad deal given the high costs to do the build. CenturyLink charges well over $500/mo for a 1Gbps Fiber+ connection without doing any buildout.
Why does the Bay have such bad housing and transit? Fiber deployment is, first and foremost, a municipal issue. Municipal governments have almost total control over the process. Some cities make it easier to build it than others: https://broadbandnow.com/research/fiber-friendly-cities. I’ve got two fiber lines to my house a few miles outside Annapolis. Verizon, which covers the vast majority of the population of Maryland, plus Comcast. I had fiber at both of my precious addresses (DC and Baltimore). My parents had fiber since 2005 back in their Virginia house, and now in their Maryland house. (They don’t even have public sewer and water.)
Whatever the big picture national issues are, there is no reason the Bay Area should be lagging even compared to the rest of the US. Rich suburbs of tightly packed single family homes are prime fiber territory. Those are the very first places Verizon wired up with FiOS 15 years ago.
Isn’t most Virginia governance at the county level?
I live in NY, and each town or city gets to negotiate franchise agreements that always favor the telcos. You have to wield a big stick with these companies, who divide and conquer.
In my cities case, in exchange for some bullshit (public access tv and filming equipment for the high school) gave Spectrum an exclusive contract for TV.
So Verizon brought FIOS to every town around us, and we’re stuck in the internet ghetto.
I was talking to a coworker earlier today (who is starting a community WISP), and it seems most of the effort/time for an ISP is about real estate negotiations, not technology.
The bay is the IT capital but it’s probably a close contender for the NIMBY capital as well. Most of the South Bay is a low density sprawling suburbia where nearly all of the homeowners are focused 100% on property value and block any development that’s not more commercial (which brings in more jobs boosting housing).
Have you seen the infrastructure of the Bay Area? It's ancient. I've had more power outages in 18 months of living here than I did in 20 years of living in San Diego(in a 70's tract. SD also has lots of old infrastructure.).
The bay area needs to be torn down and rebuilt, but the NIMBYs won't even let adequate housing be built so the chances of that happening are about zero.
Funny that you say that — I have a neighbor complaining about preserving old telephone poles, because she feels the new ones are too high, and the color stands out.
Not a Bay Area resident, but I like to buy houses in older residential areas because I like big yards and roomy interiors. Since the advent of WiFi it has been my experience that they seem to be about 10 years behind new neighborhoods. My understanding is that it's vastly easier to build new infrastructure to large developments than to bring it to older, less-dense areas.
Because cable has improved enough that no consumers really needs fiber to their homes. You can get 300-1000 Mbps cable internet now in most places and it is far more than anyone really needs. Upload bandwidth is the main limitation of cable internet but that's not a major factor for most people.
If cable internet was still stuck at 25 Mbps or similar, the pressure for fiber internet would be a lot stronger.
But cable internet got good.
The backwards and scandalous aspect of modern American internet is that the ISPs managed to roll out 1 TB data caps without being blocked by lawmakers.
This is spot on - A large majority of residential internet customers would see no difference between a 50/5 plan and a 300/30 plan. I done a significant amount of consulting in the ISP would and every time we would do mass speed increases there would be virtually no increase in aggregate traffic levels. As long as people could go to a speed test and get their subscribed speed they were happy. An HD Netflix stream takes the same amount of bandwidth no matter what your max speed is.
Most consumers aren't willing to pay the monthly cost that ISPs would be need to charge to make overbuilding fiber economically feasible. DSL providers overbuild with fiber because they can't compete with cable and the rising cost of maintenance of a copper plant. Some cable providers will do new builds with FTTH or overbuild when upgrading the existing plant is cost prohibitive. But overbuilding a working coax plant almost never makes since. Spending millions of dollars to offer higher speeds at the same price just doesn't make business sense.
Maybe it's a regional thing, but the comcast cable internet my parents have is absolute shit. They just upgraded from 300 to 600 and that did basically nothing (still get about 25 mbps most of the time and it occasionally now goes to 50 mbps in speed tests) and our video conferencing calls stutter all the god damn time which literally never happened back in my apartment that had fiber.
If we could somehow force the property market to list Maximum Internet connection speed as mandatory, thereby creating an incentive for property owner to get Fibre installed.
I live in the Bay Area suburbs and Sonic brought fiber to my home in 2 weeks with no installation charge or minimum contract and charges $40/month for it. I guess their penetration varies widely.
I live in Glen Park and I've had gigabit fibre for two years now. I don't know which ISP you spoke to but I suspect Sonic has thousands of customers in the Bay Area.
Certainly I know of at least one other guy with it myself and I cannot fathom that we're a third of all the guys with fibre in the Bay.
Fiber doesn't necessarily mean High Speed. You seem to be conflating the two.
There are plenty of areas where cable meets or exceeds the fiber bandwidth that can be sold to the home.
Anecdata, Consolidated Communications will happily sell you fiber into your business - including an MRV to convert the fiber into copper inside your property. But... the bandwidth options start out with a 10/5 Mbps line.
Hmm, there isn't much info on the tech and how the cable is physically laid.
Here is a forum topic (in French) detailing how it was attempted in France [0] ca. 20013 to bring the network up to speed. A technician explains[1] that this method is no longer used on new deployments, because of an unexpected issue: hunter's stray bullets often severed the cables. An interesting video [2] is linked in the top post.
Maybe the real innovation is that there is no need for a technician to supervise the thing. But I doubt it: it could be cheaper to pay someone to watch it than to deal with the consequences of a failure. Though I am not sure the one I linked could cross pylons, and as such could be only for HV, not MV lines.
As usual for a press release, a lot of marketing speech, light on technical details, and the infographics is bullshit, comparing top-end, almost-future fiber optics with previous gen technologies (100Mbps twisted pair vs 25Tbps fiber)
> Though I am not sure the one I linked could cross pylons, and as such could be only for HV, not MV lines.
IMO the entire point of the post is that they can now run the fiber on the Medium Voltage lines. That allows cost effectively running fiber on existing infrastructure thats much more widespread and passes closer to homes.
From the article, MV is the harder challenge for running helical cable than HV because the lines are bunched closer together, can't carry as much weight and have far more obstructions. The cool thing this new approach does is make a compact robot that can run smaller cables and handle obstacles. This is going open up fiber as an option to way more communities where it previously was not cost effective to do so.
Wait, what? If stray bullets are randomly severing cables - and doing so "often" - they should often be randomly hitting other things, in a frequency corresponding to the size and distribution of the thing(s) being hit.
Pardonnez moi for not reading the article. Je ne parle pas Francaise. But this doesn't seem like it could possibly be the reason for why this method is no longer used on new deployments.
No worries, you still get points for your effort :) (s/Francaise/français/)
This also surprised me, hence the "hard to anticipate" bit in the linked post. I guess it stems from multiple factors: high voltage lines being very long, high above ground, and crossing remote areas that might be hunting grounds. Plus, some prey might actually rest on the wires or pylons.
I should have written "lead shot" instead of "bullets", as that's a better translation. Such ammunition is commonly used for shooting birds, that can be located in front of or behind the lines. Fiber twisted around the cable also increases its cross-section.
It's actually quite likely that they often hit other things, but not human ones (stats say ~140 hunt accidents per year on average there, ~15 deadly, ~20 non-hunters). The cross section of high voltage lines present during hunts on hunting grounds is likely bigger than the human one. Plus, hunters tend not to fire at people or in their general direction, while they don't care about HV lines.
Ah, okay, that makes sense now. We call it bird shot here.
Birds are also known to perch upon wires, which is another factor that I hadn't considered before. (Actually, I wasn't even thinking about birds at all... I was only considering mammals.)
I also can't imagine that laying underground empty tubes isn't isn't cheaper over the long term. Sure, much more expensive now but they can be used with very little maintenance for decades and are much more resiliant against weather.
A big issue with aerial fiber is that the fiber experiences fairly large changes based on the environment. In areas where there is a lot of lightning, lightning strikes will cause outages. These outages are not caused by physical damage to the cable plant but rather due to very fast changes in the polarization of signals on the fiber (the huge electric field of lightning temporarily causes refractive index changes in the fiber). These index changes manifest as fast polarization rotations >1Mrad/s. This rate of change is often fast enough to make it difficult for current DSP to track. While it can be possible to track the rate of polarization change, it often comes with a trade off in nominal performance.
From the link I provided it shows the time scales. When the fiber transmission rate is 10-25Tb/s, these timescales represent non-trivial service disruptions.
From the page 4 of the link below you can see that Africa has the highest density of lightning strikes.
What does the transmission rate have to do with whether the disruption periods are non-trivial? If I read it correctly, the disruptions are on the order of milliseconds (why did you make us read through a powerpoint slideshow?).
That means not noticeable for most fiber use cases. The only use case I can imagine that would suffer from this a bit, is telerobotics (i.e. remote surgery). Unless the fiber path is so long, and the lightning storms so large that there is a continuous stream of those disruptions.
Actually the events themselves are only part of the outage. There has to be time for the link to come back up (not instantaneous). If the disruption is minor this might be a few hundred ms. However this is still long enough on OTN networks to trigger protection switching (scale is 100us). This causes traffic to be rerouted (disruptive). Additionally during a storm, it’s not uncommon for there to multiple lightning strikes in the same region. When I say same region, think the length of the state of Florida.
BTW, Florida was where this phenomenon was first observed; every summer there would be multiple outages.
The large amount of traffic on the fiber can be an issue and cause congestion and other problems (25T is a lot of traffic). Furthermore service and equipment providers have SLAs and outages can cost each money.
Sorry to hear that you felt forced to read a set of slides on the matter.
One of the problems with this new thing is that in the event of a fiber cut, restoration times will be VERY long.
For everybody who's saying "This will be great for fiber to the home!". Sorry, no. From an ISP perspective this is possibly useful for intercity fiber and medium to long haul DWDM applications. You're not going to be breaking in/out of it to add or drop circuits. It'll go point to point from electrical substation to electrical substation, at each end there will be something like a 10' cube shaped telecom equipment shelter with the optical line terminal equipment in it.
When the video mentioned a custom cable design to meet requirements, the first thing I looked for in the article was details on this:
> ...the resulting [cable] design uses G.657 200 micron fibers with a specially tailored aramid configuration and a high-strength, high-temperature, track-resistant polymer jacket to survive the requirements of this application within a small form factor.
...and the first thing that popped into my mind was a certain pain-in-the-ass-to-repair weapon subsystem which uses a certain HP/LF transmission line that also leverages a "specially tailored aramid configuration" inner jacket. Cool tech, but yikes if commercial ISPs embrace this without seriously considering sustainability implications.
200 micron fiber vs 250 for cladding diameter is extensively used in new ultra high strand count ribbon cables. The "aramid" part is just Kevlar jacketing which is also common in outside plant fiber. It sounds more exotic than it really is. Google spider ribbon fiber for some details.
The OPGW reference and your final paragraph are relevant for Transmission lines; the article implies that this robot is used on Medium Voltage Distribution lines, so not just substation to substation.
This is a cool project! What happens when that power line needs maintenance or replacement? Are you just severing the cable? Wrapping a backup line? Just doing a splice with new line and moving on?
Seems like this also produces a lot of waste over time if you wrap a power line, then have to re-do the work 30 years later (or sooner).
> Seems like this also produces a lot of waste over time if you wrap a power line, then have to re-do the work 30 years later (or sooner).
I have no specific expertise in this area. A quick Google search shows MV power lines have a life of 20-30 years, and aerial fiber lines designed for ~25 years. Facebook's goal is to get people online.
Ideally you'd be wrapping younger power lines and decommission everything at the end of its useful life. It may be viable to get people reliable high-speed internet now, and in 20 years the area is developed enough it can support even better infrastructure that they fund themselves.
Internet now might allow a merchant can list a product for sale online and post a picture, notably increasing sales. Internet in 20 years might be 1,000x as fast as it is today, and doing things we can't even imagine at the moment.
Hopefully innovation like this helps minimize the disparity in opportunity between rural and urban areas, and rich and poor countries.
This is a really innovative technology from Facebook that has a lot of potential. Props and hat tip to everyone involved in making it work.
Using existing powerlines is a double-edged sword. That infrastructure really does exist everywhere in the world. At the same time, if a powerline is cut or severed, requiring special machinery (or robotics?) to replace the fiber line is going to be a real challenge and can keep people offline for a very long time. Hopefully they're thinking of that scenario.
Power lines are increasingly buried (even MV) in cities, so it's not a good use case for this tech.
Poor connectivity in the US is a political problem, not a technical one. Google Fiber tried to solve this by getting cities to make permitting not terrible with mixed results.
> Poor connectivity in the US is a political problem, not a technical one.
Correct. Lawmakers in your state have to take steps to clear the way for broadband. Otherwise lawsuits fly when rent seekers gouge whomever has right-of-way when data services try to come through. Also, utilities have to be forced to get off the dime. If they're required to offer non-discriminatory access to poles and trenches, for instance, then they either seek to operate the broadband service themselves or come to terms with someone who can. Otherwise they behave like it's 1975 forever and squat on their power lines.
My guess is because in the US, the municipality usually does not own the power-line or pole, and in the instances the municipality does own its electrical lines and poles, it has probably already deployed fiber.
Also, having the fiber mounted lower on the pole means one can repair one or the other in isolation.
The FB method seems like it could be dangerous for repairs, but I would like to see it applied in real life, too.
Very impressive. The problem seems much more daunting than what is typically tackled with robotics. Kudos to the team. Does this mean that the electric utilities will be joining the cable tv and telephone utilities in offering internet services or do the economics only work in locales with electric transmission right-of-ways alone?
This is all middle-mile deployments, so the utility would be leasing strands to cable/telco/other entities.
There are quite a few routes built on HV power lines. Problem is you get weird problems like lightning strikes inducing magnetic field that affects polarization and other things that cause outages. E.g. https://www.ofcconference.org/getattachment/d0ec1565-ce81-48...
Deforming the coil of fiber into a U-shape is exactly what happens inside a standard box of cat-5, if you've ever opened one up. It's basically a figure-8 coil, folded into a U-shape and stuffed into the box. It pays out from the middle, just like any other figure-8 wrap.
If you want to compare with the manual process for crossing obstacles, here’s a little buggy with a gasoline motor, with the lineman hoisting it past an insulator with the help of a pulley, starting around 6:20:
I'm not clear that I'm reading this correctly, but it seems like the basic setup here was that Facebook had an idea for an innovative hardware solution, and made it happen by collaborating closely with a few experts in different related industries. This is notably different than how Google X does things, or how Facebook has previously worked with Aquila and such, and I wonder whether it'll catch on as a model.
Doesn't aerial deployment of fiber have the exact same benefits and drawbacks as aerial deployment of electricity, or anything else? In my neck of the woods (northern Europe) aerial deployment of electricity has been frowned upon for a long time: while cheaper, it is much more likely to cause outages due to environmental factors (or sabotage), so while digging down cables is more expensive, it's a lot more cost efficient in the long run because you don't have to send out technicians multiple times a year to climb poles, fell trees, patch lines etc. -- so pretty much all new deployments are dug down, and the most affected lines are converted to ground lines.
I would be interested to hear about these considerations from someone knowledgeable in electric and fiber nets, maybe fiber is somehow different?
While you are here and looking at fiber! you can run some software I've started working on.
Still very early so thought I'd add it here to see if people are interested?
The idea is to get a quick map of where the installations would be and give you a distance cost which can be adjusted based on other factors (aerial v underground, existing assets etc).
The robot is cool. The economics feel a bit off kilter for me though. With tongue in cheek, I raise the following points:
-/ Heaven forfend transferring wealth to the local laborers, to pay them to install this infrastructure. Why waste all that money by converting it into Ugandan wages when it can be spent on robots instead?
-/ Hopefully the fibre and connectivity is only leased to the Ugandans, for a more efficient return on the capital investment.
It's for safety. As explained in the article, it is not an option to shut down power to manually install the fibre because the power outage would be too long. The installation must be done with power transmission. This work is too complex for human to do safely.
The article mentions starting two robots from the same point heading in opposite directions to make that 2km instead of 1km.
But yeah, you'd have to splice it. I wonder if it would be possible to have the fibers that need splicing take a short detour down a pole to near ground level and have someone do the splice there.
Does anyone have context as to why FB finances a project like this? Is there an economic argument (ie expanding FB’s TAM by bringing more people online)? Or is there something more nuanced about better connectivity means higher quality and better converting ads? Or is this just a charitable effort?
While they have this robot and spool up there, they should look at deploying double wires for redundancy... That way you can survive multiple breaks on the line as long as each break is on a different segment.
I'm assuming that they are allowing service loops to be added to the system just for the purpose of allowing for cuts in the line to be fixed. Of course, this was just a puff piece for marketing/PR, so there's lots of details left out. However, now, instead of just the regular crew of linesmen to install/repair cables, you also have to have a robotics crew along.
I was also wondering about how they would handle these cables being replaced, or breaking due to some natural calamity, and so on. Will the locals have a way to repair the damage without the FB staff around? I'm assuming the answer is yes.
Anyone more experienced with fiber deployments--what happens when one of these km-long fibers gets damaged in a storm? Can fiber be easily spliced in the field, or does it have to be substantially replaced?
If memory serves, every splice reduces the signal to noise ratio and hence the capacity of the fiber. So you get your internet back, but not necessarily 100% of it.
The machine looks awesome, but it's so complex. It looks like it wouldn't survive two seconds in the Ugandan wilderness. Why not just zip tie the optical cable to the conductor? I bet it could move a lot faster too if it was just tacking the cable instead of this big helical movement.
i can't tell how much sarcasm to read into the zip tie comment, but no power company is going to allow the use of zip ties. the twisting of the fibre cable around the host tension cable is for safety. just attaching a cable to a suspension cable will affect how the cable behave in the wind.
that sounds like a bad installer to me. even in my neighborhood, the most recent additions to the poles have been the fiber cables, and these are even twisted around the existing bundle. it's not a tight twist, but still wrapped around though. a wrap will help prevent unwanted sagging. it's really that simple. for it not to be done just means somebody didn't do their job.
The cable is designed to be strung in the air - it has a core to help it do this as I understand it (and many others are done without a supporting cable).
