Microwave Comms Betwen Population Centers Could Be Key To Easing Internet Bottlenecks 221
itwbennett writes: Researchers from the University of Illinois at Urbana-Champaign and Duke University recently looked at the main causes of Internet latency and what it would take to achieve speed-of-light performance. The first part of the paper, titled Towards a Speed of Light Internet, is devoted to finding out where the slowdowns are coming from. They found that the bulk of the delay comes from the latency of the underlying infrastructure, which works in a multiplicative way by affecting each step in the request. The second part of the paper proposes what turns out to be a relatively cheap and potentially doable solution to bring Internet speeds close to the speed of light for the vast majority of us. The authors propose creating a network that would connect major population centers using microwave networks.
Prior art (Score:5, Insightful)
ATT had the same idea. In about 1945.
Re: Prior art (Score:5, Insightful)
Was gonna say the same thing.... or MCI, this being their entire business model, really.
Kids today! ;) Everything old is new again...
Re: (Score:2)
Was gonna say the same thing.... or MCI, this being their entire business model, really.
Kids today! ;) Everything old is new again...
I know what you mean. Next we'll be reading an article about client-server renamed to something with a C... It's a crazy world out there.
Re: (Score:2)
Too bad they've taken down most of the towers in the broadband network.
Re: (Score:3)
Too bad they've taken down most of the towers in the broadband network.
This. Many of the microwave towers in my area have been taken down in the last 5 years.
Re: (Score:3)
Many of the microwave towers in my area have been taken down in the last 5 years.
Not really surprising. My guess is the microwave towers (expensive, subject to failures from windstorms blowing radio heads out of alignment or crazy tinfoil hat people who think all RF emissions are evil, etc.) have been replaced by buried fiber optic backhaul, as fiber has become more widely available. I don't think there's any net reduction in bandwidth there.
Re: (Score:2)
No, but there might be an increase in latency...
Re:Prior art (Score:5, Informative)
For those who don't want to Google:
http://www.drgibson.com/towers... [drgibson.com]
http://www.engineeringradio.us... [engineeringradio.us]
Re:Prior art (Score:5, Interesting)
I always smile when passing old long-lines towers on the road (or seeing them on top of central office buildings in large cities). You can get an idea of the size and scope of the network at http://long-lines.net/ [long-lines.net] which has some excellent maps such as http://i.imgur.com/HI0cMJ1.jpg [imgur.com] showing the network.
Re: (Score:2)
Not dead yet, ,just turning runny, and having to be satisfied with programming
Re: (Score:3)
Actually AT&T deployed this network in the US. It was reserved as a backup communications network in case of emergency. However, it's been dismantled. The big relay towers are gone that were installed as a hub in Kansas City, and across the state of Missouri linking Kansas City with St. Louis and beyond.
Idiots (Score:5, Insightful)
Buffering and switching latency is the main source of delay, not signal latency in the copper and fiber. Microwaves would do exactly nothing to improve the switching and buffering latency. If anything they'd make it worse: light in fiber travels much further than line-of-sight microwave before it has to be regenerated with another delay.
Who peer-reviewed this paper? Did they know the first thing about networking?
Re: (Score:3)
Also every wireless signalling system I've ever seen for data tends to introduce a ton more latency with all the processing necessary for high throughput than any wired equivalents. This is even becoming a problem with high speed ethernet since past 10Gbps it's essentially RF modulation schemes again.
Re: (Score:3)
Right because a 4ms additional latency is going to make all the difference to exactly nobody but games and bottom feeding high frequency traders.
Finally (Score:4, Funny)
I can make a Hot Pocket WITH the internet! Genius.
Re: (Score:3)
Well, not so much now that The Pirate Bay is being taken down.
Re: (Score:2)
Everything old is new again (Score:4, Funny)
So....they're bringing back MCI?
Re: (Score:3)
So....they're bringing back MCI?
Not that far back.. It's WORLDCOM to you buddy....
MCI is Verizon (Score:3)
MCI is part of Verizon now. So even in 2015, the biggest competitor of AT&T is still (the parent of) MCI.
Re:MCI is Verizon (Score:4, Informative)
Verizon is the result of the merger of Bell Atlantic (who had also bought NYNEX (another Baby Bell)) and GTE. Verizon bought MCI Worldcom in 2005, which became Verizon's business division (and is also known now as "Verizon Business").
So even through all the twists and turns, the universe manages to maintain equilibrium somehow
Selective prioitization (Score:4, Insightful)
They propose shifting more latency sensitive bits to microwave links. Specifically DNS and TCP Handshakes ya know those top 2 DDOS vectors. We already have protocols to tunnel through DNS. I'm sure that will go so well.
Re: (Score:2)
There are so many ways that could be abused though - both by the ISPs and the end users.
Game server too laggy? Switch it to port 443 UDP - ISPs will think it's Skype voice and give it top priority.
Re:Selective prioitization (Score:5, Insightful)
There are so many ways that could be abused though - both by the ISPs and the end users.
Game server too laggy? Switch it to port 443 UDP - ISPs will think it's Skype voice and give it top priority.
There is a really simple solution to this. Allow users to set their own QoS rules, and the ISPs respect them, and can charge a different rate for different levels of service.
So, if you just want your SYNs prioritized it isn't a problem, and it probably won't cost you much. If you want your bittorrent traffic prioritized, that also isn't a problem, and it will cost you a fortune.
If everybody tried to ship all their mail/etc FedEx priority overnight FedEx would grind to a halt for months until they scaled up. It isn't a problem, and there are no limitations on what can be sent priority overnight, but people regulate themselves because most will not pay $70 to ship something when the $7 service that takes 2 days longer is good enough.
Re: (Score:2)
Large providers with a clue already use anycast for DNS. So assuming your ISP does a decent job peering you should never see a google DNS request going from NYC to Chicago outside of a failure of the DNS servers in NYC.
I don't understand.. (Score:4, Insightful)
.. why we would want to use microwaves for this. Fiber is shielded, and capable of higher throughput. While I can understand using microwaves to communicate with satellites, I don't see why we would use them for communications between two population centers.
This might just be my dislike of wireless in general, but I don't see how this could solve latency issues...
Re:I don't understand.. (Score:4, Informative)
Speed of light in fibre is about two-thirds that of vacuum.
Re: (Score:3, Informative)
Also the fact that microwave towers can be placed more directly than trying to lay a fiber directly between population centers. Most fiber runs along highways/railroads which have other constraints than must be straight between two points.
Re: (Score:2)
Microwave is also line of sight, which means a tower and amplifier every 10 miles minimum. You know that problem with a curved planet where you microwave is straight line.
Re: (Score:2)
And if ~60GHz bands are used, it stops working when it rains...(oxygen absorption).
Re:I don't understand.. (Score:5, Funny)
So remove the air from the fiber. Make it a vacuum. God do I have to think of everything?!?!
Re:I don't understand.. (Score:4, Funny)
And then sell the "oxygen free" fiber cables to Hi-Fi nuts who want to improve the quality of their streaming audio.... Profit!
Re: (Score:2)
Monster Cable's lawyers called. They'd like to have a word with you regarding their patented business process...
Re: (Score:2)
Communication fibers are solid, not hollow. Making them hollow would would end up completely defeating total internal reflection [wikipedia.org] that fiber depends on to actually send over a distance.
Re: (Score:2)
3 minutes after your post, someone posted something that totally disagrees with what you post.
http://tech.slashdot.org/comme... [slashdot.org]
It turns out that hollow fibers transmit faster than solid.
Re: (Score:3)
I had a similar idea about Zeppelins. Hydrogen is dangerous, Helium is expensive, so why not just pull a vacuum in the lift cells? Empty space is much lighter than helium, just think of the buoyancy! Everybody is an idiot, except for me.
Re: (Score:3)
That was first proposed in the 17th century by monk Francesco Lana de Terzi.
The elasticity divided by square of the density would have to be about 4.5 that of diamond. Such a material might be made someday
Re: (Score:2)
Yes, Except for hollow optical fibers. Dunning-Kruger yourself.
http://www.engadget.com/2013/0... [engadget.com]
Re: (Score:2)
Whoosh.
Re: (Score:2)
Re: (Score:3)
The speed of photons is not a meaningful part of that latency.
It is if you're trying to siphon money from the stock market without adding value. [extremetech.com]
For the general internet, though, it is useless.
Re: (Score:2)
It would be much faster to use microwaves to go from city to city rather than use satellites. It's much faster to from one point to another rather than point to satellite to other point.
Latency is caused by storing packets (Score:2)
Re: (Score:3)
You can make money with arbitrage, and do society a benefit at the same time.
Re: (Score:2)
Re: (Score:2)
Rain fade. (Score:2, Informative)
Re: (Score:2)
And multimode fiber for last 100'.
Re: (Score:2)
You can also use singlemode for short distances, you just have to put your light meter on and attenuate the signal down to avoid burning-out the receiving end. Most providers that use singlemode tend to use it exclusively so they don't have to carry multiple sizes of fiber patch cords, and while the transceivers are
Re: (Score:2)
Sure, but I was going to write last mile & knew that was wrong. Meh, 100' is a reasonable start.
Re:Rain fade. (Score:4, Funny)
Microwave networks are extremely susceptible to rain fade, and as such are not a good choice for important data links like these would be. We already have a technology which allows signals to travel at the speed of light and is immune to weather, solar radiation, and nearly anything else short of a major earthquake. It's called single mode fiber optic cable.
I didn't know a hung-over backhoe operator was considered in the same class as a major earthquake.
What exactly has caused your last three fiber outages? Chances are it was a human behind a stick or wheel, and not Mother Nature.
Re: (Score:2)
sorry pal, speed of light in fiber is much slower than in air, roughly c divided by cladding index which might be 1.4 or so
Re: (Score:2)
oh? what bottlenecks would NOT be common to long distance by fiber compared to microwave in air
Re: (Score:2)
Depends on what you mean by "speed of light", normally when people say "speed of light" with no other qualifiers they mean "speed of light in vacuum".
The speed of light in air is near as damnit the same as in vacuum. The speed of light in fiber is somewhat slower. So if latency is king then microwave wins. The reaearchers posit that by pushing the most latency sensitive packets onto a low latency network they can improve overall performace.
The big issues I see in practice would be
1: getting anyone to pay fo
These wouldn't be the microwave comms... (Score:4, Interesting)
... that were slowly dismantled in the 90s because fibre optic was supposedly better would it?
You have to laugh. Another generation comes along and re-invents the wheel. Again.
Re: (Score:2)
Re: (Score:2)
Yes, this. Read the article, they are talking about 400Mbps microwave links. That's a drop in the bucket compared with fiber bandwidth. This paper is all about latency above all else. First, I remain unconvinced that RF links are really different than fiber for latency. Second, I'm unconvinced very many people care about the difference of ~5ms in latency (using their numbers) and would consider bandwidth much more important. They point out one particular use case for low-latency, low-bandwidth links like th
Rain rain go away (Score:3, Interesting)
Some microwave frequencies are sensitive to the weather.
I'm not sure if there are any that are weather-insensitive to be useful in a thunderstorm, snowstorm, or in heavy low-lying clouds/foggy conditions.
Re:Rain rain go away (Score:5, Interesting)
Re: (Score:2)
Yep. Microwave fade. It's inherent in ultra high wavelength transmissions. That said, we have modulation techniques today that have effective error correction, unlike back in the day when this was all done with FM and you ended up losing data when you had poor conditions. Think about how your mobile phone gets massive bandwidth and reliability out of OFDMA (LTE's downlink technology) under much worse propagation conditions in city canyons, and scale that to microwave frequencies.
That said, given the bandwid
Fiber is fast! (Score:5, Insightful)
Fiber is amply fast.
The bottleneck is the cavalier attitude of web designers to network resources. You do not need to load 25 different URLs (DNS lookups, plus autoplay video and all the usual clickbait junk) to show me a weather forecast. Or a Slashdot article, for that matter...
...laura
Re:Fiber is fast! (Score:4, Insightful)
And a dozen javascript libraries for stupid shit like mouse-overs that should be done in CSS anyway, or high resolution background images that are 2MB JPEG downloads that use over 6MB of RAM each once decompressed. Backgrounds aren't meant to be high-resolution, crisp and detailed. Learn to use background-size: cover, it works well even with lower resolution images because stretching will blur them a bit, making the compression artifacts even less noticable.
Re: (Score:2, Informative)
Yep, nothing makes you realize why modern websites are so slow like the first time you install NoScript. I never knew before then how many websites were having me download half the damn internet over and over again.
Re: (Score:2)
background-size: cover bugs out on iOS (Score:4, Informative)
caniuse.com's background-size chart [caniuse.com] claims that Safari for iOS has defects in its handling of background-size: cover.
Re: (Score:2, Funny)
Sure, let's solve this problem by ... uh .. what, legislating HTML practices? Brilliant.
Re: (Score:2, Interesting)
Google results preferring quickly loading pages.
Re: (Score:2)
We could make fun of every website of inep design.
Re: (Score:2)
Re: (Score:3)
Define "fast." This is apparently not about download speed but about latency. The idea is apparently to keep the majority of traffic that doesn't care about latency on fiber and move only that which does to a microwave network. (How do you do that? They didn't say.)
I'm not sure why they think latency is a big issue. Latency simply isn't a concern for the vast majority of Internet applications. They admit as much in the article and claim the majority of traffic would remain on fiber links.
So what's left that
Re: (Score:3)
Latency is a big issue for the web. You start off with a round trip for the dns query, then another couple for the TCP connection (and more if it's SSL), then many more for TCP to figure out the channel capacity and come up to full speed. Our broadband connections are now "long and fat" enough that latency not bandwidth is the limiting factor in how fast a request can be completed in most cases*. Multiply crappy web design that requires large numbers of requests, possiblly from different servers and possibl
Having worked at weather.com for 10+ yrs (Score:2, Funny)
You made my day! :)))
You have no idea how many wars were fought over: "WHY IS THE SITE SO SLOW?!? Google's fast! Why can't you make us fast like Google?!?"
Yeah, that was a real mystery to all of us who weren't in product or ads...
Re: (Score:2, Interesting)
Re: (Score:2)
You are obviously not the target demographic for most sights.
perhaps you should try a site that targets low bandwidth users ... Sclog!
City/Nation Wide Mesh Networks (Score:2, Interesting)
Mesh networks, peer-to-peer between nodes (homes/businesses,etc), would be useful for offloading non-time critical applications such as file transfers and open up opportunities for local delivery of online services that eliminate the need to send traffic through major choke points.
Where are the 'slowdowns' ? (Score:3)
Re: (Score:3)
Not exactly. The slowdowns happen in switches, routers etc that have a heap of software loaded on them that do a lot of other things besides switching and routing such as port duplication (allowing the NSA and other taps), VLAN, L3/4 packet inspection etc etc all of which are done on low-power early 2000's devices (remember how "fast" embedded chips were back then) with crappy, H1B-written software and features bolted on top over the last 2 decades.
Using microwave links won't help either, it's not the mediu
Free lunch! (Score:2)
Bandwidth is not infinite, and due to that, it won't scale up.
This is one of those pie in the sky solutions that simply doesn't scale, and even if it could, would manage to only work by stealing other spectrum, and in the end runs up against the physical fact that there isn't any more spectrum being made: Here's the spectrum chart reality check:
http://www.ntia.doc.gov/files/... [doc.gov]
Pick who you are going to kick off.
Re: (Score:2)
>Bandwidth is not infinite
Well... in practical finite-but-very-large terms, it's really a matter of how much money you're willing to throw at it.
You can put up a lot of microwave transmitters, and so long as your receiver is designed to be able to pick out the sources - much like a camera can have more than one element registering 'red' - you can use the same frequency range for all of them.
As long as neither the receiver or transmitter are moving significantly, this isn't technologically impossible.
Re: (Score:3)
You can put up a lot of microwave transmitters, and so long as your receiver is designed to be able to pick out the sources - much like a camera can have more than one element registering 'red' - you can use the same frequency range for all of them.
As long as neither the receiver or transmitter are moving significantly, this isn't technologically impossible.
Talk to me about the noise floor.
Re: (Score:2)
Care to elaborate what "Fixed" means in that chart?
Also, wow.
Re: (Score:2)
Care to elaborate what "Fixed" means in that chart?
Fixed is referring to fixed station operation. It might be a safety purpose, such as maritime, and cell service towers and microwave. Generally shared with mobiles of course. Most of the time we will look at the other purpose, like "Maritime mobile", and then can figure out the main purpose of the band. There must be some historical meaning to it, because otherwise, it seems kind of redundant.
Also, wow.
Especially since the parts we are interested in for computing use - not just microwave, is roughly from 2 GHz and u
Re: (Score:2)
Thanks! Very interesting
Why not lasers? (Score:2)
Lasers would offer the same line-of-sight links, but with much more bandwidth. It could be used with existing cable as a backup for bad weather, just like the microwave proposal.
Re:Why not lasers? (Score:4, Interesting)
Not even lasers. Hackers have been doing this with freaking LED's for long range networking.
http://en.wikipedia.org/wiki/R... [wikipedia.org]
Short of really massive weather conditions they are reliable as hell and dont require clearing all the trees out of the frenel area in front of the dish.
Re: (Score:3)
Yeah, then we could point the laser at a glass tube to keep the rain out. Then we could make that tube flexible so it can go around the curve of the earth and we don't need to build a tower. Hey what do you know, we just made fiber.
I had microwave Internet 15 years ago... (Score:4, Informative)
In Lousiville, CO, I lived in one of the few neighborhoods that was skipped over for broadband in 1999. Sprint setup a microwave service that filled in the gap. Bandwidth was awesome - I was getting 10-30 MBs regularly. The downside was the latency - 100 ms ping times were the norm. I remember trying to play Duke Nuke 'Em with friends and having the unfair "advantage" of disappearing regularly when my client didn't ping back in time. Being line-of-site, there were also issues with trees occasionally swaying in front of the dish (a pizza box attached to my roof) and snow blocking the signal.
As others have pointed out, microwave Internet isn't something new and, unfortunately, in the real world isn't a perfect solution.
-Chris
Does latency really matter? (Score:2)
Does anyone besides gamers and high frequency traders care about latency? I manage a server farm on the other side of the country, and latency is not an issue at all for interactive use whether typing at the command line using ssh or using RDP to connect to a Windows server. For the general internet user, I don't see much utility in cutting round-trip latency in half from the current 60ms I'm getting now to the 30 ms a speed of light connection would give. (though there'd be additional latency from all of
Re: (Score:2)
If you play multiplayer online games, latency is a big issue.
If you talk to people over the internet, latency is an issue. Like, you say something in Skype. The person at the other end hears it and replies. By the time you hear the reply, a regime change has taken place and there's a new president in power. Currently internet video chatting over long distances is an unpleasant experience due to the lag.
Re: (Score:2)
If you talk to people over the internet, latency is an issue. Like, you say something in Skype. The person at the other end hears it and replies. By the time you hear the reply, a regime change has taken place and there's a new president in power. Currently internet video chatting over long distances is an unpleasant experience due to the lag.
http://www.voip-info.org/wiki/... [voip-info.org]
Callers usually notice roundtrip voice delays of 250ms or more. ITU-T G.114 recommends a maximum of a 150 ms one-way latency. Since this includes the entire voice path, part of which may be on the public Internet, your own network should have transit latencies of considerably less than 150 ms
I'm already getting cross country ping times of 65ms (round trip), so to be compliant with ITU-T G.114, my codec has 235 ms to do its work. I regularly talk on the phone with colleagues on the other side of the country using a VOIP system hosted here, and haven't noticed any latency problem. Even video calls using our Polycom have good latency (but not great, there's still a noticable lag, even when connected to local users)
I find voice latency on cell phones (even local c
Additional bonus? (Score:2)
Perhaps we will have less snow to shovel then.
I have an idea (Score:4, Interesting)
Wonderful idea (Score:3)
Canada had a microwave network across the nation by the end of the 1950s.
https://www.historicacanada.ca... [historicacanada.ca]
Why? (Score:4, Interesting)
Just use the frigging dark fiber that is already running between them.
Just a bunch of (Score:2)
Garage Door Openers (Score:2)
Never Happen (Score:3)
They will continue business as usual overcharging us for their wares and delivering nothing.
Only because they're looking at point-to-point (Score:2)
The only reason they're thinking it has less latency is because they're only considering a point-to-point case.
Once you wire that up to the various hubs and routers of the underlying DSL or cable infrastructure, or try to send traffic to somewhere other than the end-point the microwave is connected to, your latency rapidly goes to shit.
Add in the fact that fiber can transmit over longer distances, the weather-induced degradation of most microwave links, the fact that the whole link drops when a semi-tr
From the very title: "Towards a Speed of Light..." (Score:2)
For faster Internet they clearly wants more bits to move as photons, at the speed of light through fibre. Nothing is faster (latency, throughput, bandwidth), and all the nearby alternatives including microwave as more expensive and less reliable.
The organizations that had microwave towers for communications, namely telecommunication companies and media broadcasters, have long since migrated to a) satellite or b) fibre for their primary connections. The only microwave links that I know of locally (~100km) ar
Light is still the speed limit. (Score:4, Funny)
I remember getting a request for a cluster that was proposed to be split between a midwest USA site and London. Conversation was something like this (not exact numbers, but I did do the math at the time):
PHB: We need less than 50ms latency.
Me: Can't be done. We're at around 120ms right now with 10ms jitter using VPN.
PHB: What about MPLS?
Me: That might get us to 115ms with 5ms jitter.
PHB: Well, we have to come up with a solution. What else can we do?
Me: Slap Einstein? This is a physics problem, not an IT problem.
PHB: This is OUR problem to solve.
Me: Ok, if we buy our own glass, lay it in a straight line between us and London, which also includes some sort of housing for it that I don't know exists that would prevent issues with freezing/melting/icebergs, we'd end up with 72 ms.
PHB: So there really isn't anything we can do...*starts walking away*
Me: Hold on! I have another idea! We can tunnel through the earth, and skip the water issue if we can come up with a new type of shielding that can withstand tectonic forces and heat. That will allow us to get to 55 ms since we're not dealing with the curvature of the earth! Will that work?