Forward-thinking technologists, including me, have been predicting for some time that adaptive mesh networking would be the doom of the telecomms-carrier and broadband oligopoly. Now comes a scientist from Australia with an idea so diabolically clever that I’m annoyed with myself for not thinking of it sooner: put the mesh networking in smartphones!
For those who came in late, “adaptive mesh network” is shorthand for a way to do Internet everywhere by building cheap wireless Internet routers that communicate over unlicensed radio spectrum and self-organize by handshaking with their neighbors. Such a network can relatively easily be engineered to autoconfigure and heal itself against point failures.
In Dr. Gardner’s scheme, smartphones equipped with adaptive-mesh hardware and software take the place of the dedicated mesh nodes everyone else has been imagining. Here’s what makes this diabolically clever:
- By pitching it as a backup in case a power outage or disaster takes out cellphone towers, Dr. Gardner makes it an easy political sell and hard for the carriers to oppose. Regulators might even mandate it as a way to solve the 911 problem!
- Smartphones already have to have most of this capability anyway; adding one more frequency band and adaptive-routing software is no big deal.
- The scheme amortizes the cost of the mesh support across the smartphone vendors’ huge production runs, lowering cost per unit way past anything achievable for dedicated mesh nodes.
- Very clever attack on the power-distribution problem; people will solve it every day because they want to keep their individual phones running!
- This mesh would be self-deploying; instead of fixed node locations, it goes where the users are.
There would still be a place for fixed-location mesh nodes as a backhaul network to increase aggregate bandwidth and service reliability. And they’d get much cheaper, because cellphone production volumes would pull the cost of the repeater hardware down to zip. People would buy sixpacks of booster nodes at supermarkets; banks would give them away like they do calculators now.
This idea is beautiful. It is made of goodness and winnitude. It’s technically feasible, should be politically viable, and gets on the right side of economics of production scale. Hey, Google! Want to really undermine the bandwidth monopolists? Throw some funding at this guy now – better yet, hire him and put the mesh support into Android!
Every hacker everywhere should be annoyed he didn’t think of this.
The one laptop per child people did this, right?
It doesn’t help if you are out of Wifi range of your nearest neighbor.
What does the extra traffic do to your battery life?
So it’s like extending phone coverage via P2P?
ESR says: More like extending Internet coverage via P2P. Then you do voice over that.
I know there was a prototype wifi router that tried to do the same thing with Internet connections, but I can’t remember if it was ever put into production or what it was called.
Damn it! I feel stupider now.
Actually, the only bad thing I can see about this is that it would work really, really well in densely populated areas, but not so well in more rural areas. Modern digital cell phone radios are fairly low-powered devices, and the WiFi radios in them are even lower-powered: the WiFi tethering on the EVO 4G, for instance, feature only works for about 30 feet.[*]
In rural areas, you’re either going to be tied to the cellphone network or another available broadband connection.
Rural areas is the one thing mesh networks always fail to address. It matters for me personally, because while I live in a fairly densely populated area, about 2 miles directly east of me there are horse ranches, cattle ranches and blueberry farms.
[*] I have yet to test this, however, this is what the manufacturer and Sprint both claim.
That was my first thought — do I really want my cellphone viable life reduced from 20 hours to 1 hour just to let some (insert favorite antisocial descriptive mild expletive here) watch a movie?
I think it’s a great idea, but we’re still dealing with the social implications of people who talk on the phone while in line at the grocery, driving, etc. Now, they’d be wasting other peoples’ batteries (and bandwidth) as well. I almost think the idea would work better if we got back to the grandparent’s party-line mentality. This might require actually having a “party line” or, in modern parlance, if you want to ship your bits through my phone then (a) they better be unencrypted so I can see if you’re wasting my battery on drivel; and (b) they better be signed so I know who to cut off…
My second thought was — the article claimed they needed some funding. If they have anything worthwhile, you’d think google would be all over that…
BTW, Sprint sells some iDEN (formerly Nextel) network phones which work like walkie talkies where there is no service.
Sometimes old but good technology is best.
Quasi-related: the smartphone as primary computing device seems to be a bit closer to reality. Wireless PCIe.
That solves my main objection: insufficient graphics performance. Nice.
>Quasi-related: the smartphone as primary computing device seems to be a bit closer to reality. Wireless PCIe.
>That solves my main objection: insufficient graphics performance. Nice.
I knew it! External discrete graphics acceleration for your phone!
>In rural areas, youâ€™re either going to be tied to the cellphone network or another available broadband connection.
And that’s why the second-order effect of making dedicated mesh nodes really cheap matters. Having spotty coverage in your rural area? Give some nodes that plug into wall sockets to your neighbors; helps you, helps them.
It would also be foolish to assume that the repeater spacing will have to be WiFi like. Consider what such a network would look like if it started with WiMax base ranges, then added dirt-cheap Pringle-can antennas.
Did you hear about the mesh potato?
People in the 3rd world need the democratized grid more than us in the west.
Wow… that is brilliant, yet simple and elegant!
The only issues I see:
1) battery requirement — if the phone is constantly routing data, it will draw more power. This will likely get fixed as battery tech gets better.
2) low-density areas. Sure, in urban and suburban areas, the distribution of phones will be good enough to provide for this case. What about more rural areas where people could be a mile or more apart? Yeah, I know in some they have zero signal anyway, but in others, mostly due to government subsidy, the carriers have towers available.
I dimly recall Bob Cringely pitching something similar about 5-6 years back, but installed in new cars. Like most of Bob’s pitches, it went nowhere.
Another thing that needs to be done, and probably will be in 5 years: no more dedicated voice networks. All cell phones should be IP devices that use VoIP. This will make the voice calls more network agnostic, and will be able to automagically route over LTE/WiFi/mesh network/whatever.
We can be confident that the despised oligopolies won’t engineer incompatibilities to render all of these wonderful possibilities moot.
Also, look for the FCC to magically redesignate everything in such a way as to put a mailed fist around the tender parts of all these gadgets.
And your enthusiasm.
Crawling out from under my wet blanket, let’s get to work on doing this. Up the oligarchs.
WiMax has to get more affordable first, but, yes, I see it.
Nope. That’s a non-starter. Margins are tight in the car business (even for the Japanese) and anything that increases the cost of the car without providing some benefit for the car maker will be immediately rejected. Why do you think the car makers get all up in arms whenever they talk about increasing CAFE standards or requiring some new piece of equipment? They’ve had enough regulation to last several lifetimes as it is.
If we’re talking about twisting telcom arms to get this in smartphones, why not? The hardware is mostly there already.
I don’t think that battery life will be too much of a problem, at least in cities. In a crowded area, there will be lots of phones available to route through, and yours will only get hit a small part of the time.
(This assumes that everyone isn’t yakking at the same time. In that case, even the regular cell networks have trouble. Mesh networking is still going to be vulnerable to overload in emergencies.)
I’m not sure this will scale well over large distances. Imagine a cross country phone call over this network. You will have to go through hundreds if not thousands hops. For low latency activity, such as voice traffic, it would be a nightmare. You could, of course, have points of transfer into a regular fiber network which can then take care of long haul. But it would remove a lot of the “ad-hoc” aspect of it.
I think this concept was used in Charles Stross’s novel Halting State.
Exactly why it’s a perfect match. I was referring to Cringely’s idea of putting nodes in cars, which is a non-starter for the reasons I listed.
> What does the extra traffic do to your battery life?
It’s going to pistol-whip hell out of it.
And that’s probably the number one reason it’s troublesome. You get me 10 or 20 times as much battery life per weight as we have now, and it’ll fly. Til then, probably not. Brilliant phones are hard pressed to make it through the day now.
“We can be confident that the despised oligopolies wonâ€™t engineer incompatibilities to render all of these wonderful possibilities moot.”
My first impression is that if allowed to proceed as Eric suggests, wouldn’t this potentially foreclose (or at least greatly diminish) the possibility of mass media/Internet censorship?
The Cybiko had almost exactly this functionality back in 2000. It would be a huge boon for connectivity at events like music festivals, where traffic is mostly internal to the gathering.
Damn! And now I’m tempted to buy an android, just so that if/when this happens, I’m ready…
The battery life thing will probably be moot in suburban areas – many if not most people and businesses will install wall-plugging nodes to save their own battery life, improve speed/latency, whatever.
Once this sort of network becomes common, I could envision carriers being forced, by one mechanism or another, to turn their towers into nodes, improving the battery/latency/rural-area problem further.
>The battery life thing will probably be moot in suburban areas â€“ many if not most people and businesses will install wall-plugging nodes to save their own battery life, improve speed/latency, whatever.
This is right. Everybody who’s thinking this won’t work because battery life is limited is forgetting the second-order effects of pulling down the cost of the repeater hardware.
A bit off-topic, but here’s sonething about the new Droid, http://www.mobilecrunch.com/2010/07/14/droid-x-actually-self-destructs-if-you-try-to-mod-it/
If you want to be on the bleeding edge of this phenomenon, you might get involved with the Open Mesh project:
Open Source hw and fw, cheap, robust. I provide internet service to several of my neighbors this way.
> Every hacker everywhere should be annoyed he didnâ€™t think of this.
Hmm, I thought of this years ago but rejected it as a clumsy idea. Too hard on our already-limited phone battery life, bandwidth, and available CPU power. Too complicated to build and route through a constantly-morphing network. Too many network hops to reach anywhere important.
I mean, seriously, your smartphone is slow enough today. Do we really want to see what happens when ten nearby strangers are downloading anime porn through your uplink?
Of course, it’ll happen eventually, on the scale of decades. It’ll be too useful when your phone has a poor/non-existent connection to the real network to find a backup route through a nearby peer. But by the time this morph networking is deployed we’ll probably also have a cell tower network or satellite network that’s much more reliable than it is today.
Also this will be useful for P2P applications like gaming. Much better to send a wifi packet straight your buddy playing Mario Kart next to you on the train, than to route through a cell tower and back.
The US military has put a fair bit of research into this sort of architecture for battlefield communication. As far as I know, they haven’t deployed any of it, and the COTS systems may beat them to it, but there might be some useful public papers from those programs…
The main problem this would face would be the fact that there are TWO powerful lobbies to go after: we can’t forget that the media has a big interest in keeping ISPs around, so they can pump out DMCA notices without regards to the actual situation. When the carriers of “pirate” traffic are so diverse and changing by the very nature of the system, individual liability becomes difficult to establish precedent for. Not everyone with a smartphone is going to have a legal office, and fewer are going to keep logs of incriminating evidence.
The key is to get this deployed so quickly that none of the legal incumbents can react quickly enough to squash it.
Once deployed, once I can actually do low-key web browsing over the local mesh, it’s going to very hard to stuff that genie back in the bottle.
Jeremy Bowers Says:
“The key is to get this deployed so quickly that none of the legal incumbents can react quickly enough to squash it.”
Yeah! That’s the ticket. Force it on people before they know what happened! I love how people look at something they believe to be a good idea, then want it forced on everyone, often with “funding”–you know, the money that just grows on trees. Why don’t we do the same for the trucking industry? Think about how many big trucks we could get off the road if everyone took a package for delivery with them when they drove somewhere. I don’t use Skype (Stickam rules!) for the very reason that I want to control what passes through my computer. Why would I want your porn and spam to take its place? So like Patrick Maupin says:
“…if you want to ship your bits through my phone then (a) they better be unencrypted so I can see if youâ€™re wasting my battery on drivel; and (b) they better be signed so I know who to cut offâ€¦”
Ditto for the “free” node the bank is going to give me.
>Force it on people before they know what happened!
I don’t think anyone here would object to a having a software switch in the phone that says “don’t route peer-to-peer traffic” so you could opt out. The point here isn’t to force anything on anyone, it’s to give users a way out from under the thumb of the telecomms monopolists.
# esr Says:
“I donâ€™t think anyone here would object to a having a software switch in the phone that says â€œdonâ€™t route peer-to-peer trafficâ€ so you could opt out. The point here isnâ€™t to force anything on anyone, itâ€™s to give users a way out from under the thumb of the telecomms monopolists.”
Question is, how many here would object to an opt in policy? Why make people opt out of something when it is just as easy to give them the opportunity to opt in? No matter how you slice it, someone is tasked with supporting whatever infrastructure is used to get calls (and data) from sender to receiver. Having to opt out means that unless I choose otherwise I am basically an employee of the company that maintains that infrastructure. If they pay me for performing that service, it’s one thing. Turning people into unpaid labor unless they take some action is another. Everything being discussed here and in the article can be done voluntarily, without making cell phones part of a P2P network. Every existing cell tower could become a fixed node of the network. Developers and businesses could put them on the exterior of their buildings. If people wanted to they could buy as many “sixpacks of booster nodes” they wanted, or get the free node from the bank and plug them into an outlet at their business or home to achieve the same goal. And you could also provide incentives for people to opt in, rather than having to opt out–price breaks on phone service. If it’s such a cool idea then there should be no problem getting people to *want* to become a part of the mesh. One person here has stated that they provide free Internet to their neighbors with nodes they installed. I bet 90%+ of the others here have their wireless networks secured and buttoned down as tight as possible.
I feel uniquely positioned to contribute to this discussion, having spent 5 years building the technology behind a startup company called “MeshNetworks, Inc.” and having spent another 5 years at the company that acquired us, called “Motorola.” (Check the redirect of our old web site: http://www.meshnetworks.com.) We built a fairly successful line of outdoor wireless mesh networking gear in both licensed and unlicensed bands for both public safety and municipal data systems, using both proprietary and standard (Wi-Fi) radios. We turned that proprietary technology into a software library that could be dropped into any product in a few weeks, including cell phones. We spearheaded the effort to standardize our approach inside IEEE, in the meshing standard that will be called 802.11s, if it ever makes it out of committee. If anybody was positioned to pull this off, it was us.
And I’m here to burst your bubble.
Trust me, I’ve drunk the kool-aid. I spent years studying at the feet of PhDs and professors in their ivory tower classrooms, listening to them describe the elegant beauty of ad-hoc mesh networks. This stuff was old news to me by the time Cringely published his article around 2002-ish. And I’ve spent a total of 10 years refining the protocols and systems. I think it’s fair to say I’ve seen how far the technology can go, and for many use cases, it’s really terrific stuff. Awesome. Fantastic, even. I’m still the world’s biggest supporter of mesh networking, even though I left Moto last year.
However, w.r.t. purely ad-hoc meshing between consumer devices, the problem with this utopia of “Mesh everything!” can be found in the word “unlicensed.” All the unlicensed bands are unlicensed because, from a physics perspective, they suck (resonance of water, etc.). And because there aren’t that many bands/channels available, they’re chock-a-block full of crap. Sure, your spiffy 802.11n AP gets you ~300 Mbps in an open green field with no civilization for miles. Carrying a voice call is easy! But in a tree-lined downtown where everybody has a microwave oven, a baby monitor, a portable phone, and a few dozen Wi-Fi/Bluetooth devices in their house, the band starts to break down quickly, even at 5 GHz. And 5 GHz really sucks at penetration. That means you need more energy and high-gain antennas to pierce through the noise and walls. That means that devices stop Tx’ing at 300 Mbps and drop down to lower data rates in order to overcome the noise through coding gain. And both of those factors mean that the data that does get sent takes up more time on the air, and consumes a larger radius. Think about those two factors: As you might imagine, they begin a *very* vicious cycle, especially once your Wi-Fi signal leaves the protective Faraday cage that is your house or the corner Starbucks.
In addition, although it was amazingly cool to build a protocol that could sustain a dozen or more hops and keep the mesh functional when nodes were moving at 90 MPH (and we did it!), in reality, nobody will tolerate that kind of latency for full-duplex voice, and the out-of-order packet arrivals from the dynamic route changes are enough to really screw with TCP. So in the real world, all the successful mesh deployments limit their wireless hops to just 2 or 3, and seed the mesh with fixed, wire-backhauled infrastructure with big antennas and 120V feeds. We had some success with vehicle-mounted and handheld nodes, but beyond some very niche scenarios (er, think “black hats” and “desert camo” and other people who can freely ignore FCC power limits), they usually don’t work out as well.
I could go on for days about the real-world issues and how we solved or didn’t solve them (security, latency, jitter, doppler, multipath, interferers, split horizon, hidden-node, multicast/broadcast distribution…), but IMHO, here’s what it would really take to implement your mobile phone mesh utopia: A licensed band with a mandatory channel fairness protocol and mandatory dynamic output power control algorithm, opened up by the licensor to allow sub-licensed consumer gear to operate. The possibilities are amazing, but the chance of it actually happening, I’m afraid, is near nil. I hope I’m wrong.
>The possibilities are amazing, but the chance of it actually happening, Iâ€™m afraid, is near nil. I hope Iâ€™m wrong.
Would these problems be effectively addresses by opening up more spectrum to unlicensed use, or do you see the problems as more fundamental than that?
The reason I’m asking is because if more spectrum in different places (e.g. with better non-line-of-sight propagation) can solve the bottleneck, then the problem is subject to political attack by a big player like Google.
I had exactly the same reaction as esr: It must be obvious to all the technical people involved that this won’t stop at being just a “backup system for emergencies”.
> itâ€™s to give users a way out from under the thumb of the telecomms monopolists.
Mesh networks don’t solve the need for long-haul network backbones.
You don’t want to browse the web over a mesh network built mostly out of people’s cell phones. Too much latency. Even a super high speed, super reliable mesh network would have too many hops to comfortably connect to web sites across the country or even across town.
Just like with cars, the only sane way to travel long distance is to get on the highway as soon as possible. And the on-ramps to the internet’s highways are tool booths controlled by the telecoms. How does putting a public wifi hub in every smartphone solve that problem?
“Would these problems be effectively addresses by opening up more spectrum to unlicensed use, or do you see the problems as more fundamental than that?”
Question is, from who do you take the bandwidth? Virtually all the broadband providers are now lobbying to take it from the UHF band, claiming that it is not currently being used “efficiently”. Years ago there was a commercial warning against the “monster” of cable TV (http://www.youtube.com/watch?v=FcYL2PW_fjE). That commercial has been resurrected because of the drive to take bandwidth from the UHF part of the spectrum. Do we eliminate free, broadcast TV (all broadcast TV is now UHF) to feed the desire for Internet access wherever we happen to be standing?
> Do we eliminate free, broadcast TV (all broadcast TV is now UHF) to feed the desire for
> Internet access wherever we happen to be standing?
Considering the Internet actually achieves what TV was originally envisioned and intended for all, I’d answer ‘yes’ to that.
If mesh networks would work for cell phones, they would work even better for home networks. You’d only need a few houses per neighborhood and a few buildings per office park to provide public mesh-fi nodes, and the whole industrialized world could stay connected through the mesh.
Why hasn’t this happened yet?
(One reason is that current wi-fi tech has a range that is an order of magnitude too short. A modern G or N router can barely serve one house reliably. But that doesn’t sound too hard to solve.)
>>Do we eliminate free, broadcast TV (all broadcast TV is now UHF) to feed the desire for Internet access wherever we happen to be standing?
All TV shows should be on the internet in the first place and be easily be accessed by anybody in the world. Of course, it will be supported by advertisements and whatnot.
As of right now, it is still not easy to watch TV shows on the internet like people flipping channels.
> Would these problems be effectively addresses by opening up more spectrum to unlicensed use, or do you see the problems as more fundamental than that?
Full disclosure: I am a software and protocols geek, not a modem/radio expert. Anything I say about modems and spectrum should be taken with an appropriate grain of salt. But like I hinted before, I think the big issue with unlicensed is that there’s nobody dictating how to “play fair” in the spectrum.
Sure, FCC puts raw dBm caps on the bands, but to make a system efficient enough to support The Vision, I really think you’d need a mandatory power control system, so that two near nodes talking over 20 feet won’t pollute the spectrum for a quarter-mile. I also think you need a channel fairness system, so that nodes gracefully relinquish the band and let other nodes access it. Technically, those are both solvable problems (although I admit, dynamic power control is a bitch, and something we never really got right).
In pure unlicensed bands, though, there’s absolutely *no incentive* for a device manufacturer (or a user) to “play fair” in either the power or time dimension. Every Wi-Fi manufacturer runs their designs at the bleeding edge of output power (usually butting up against the output mask and distortion in their amplifier), since power benefits both range (duh) and throughput (by overcoming noise, thus allowing higher modulation rates). No user ever turns down the power. And sure, Wi-Fi has QoS and inter-frame spacing standards to relinquish the channel, but cordless phones and microwave ovens don’t much care for your Wi-Fi standard. Even standards-compliant Wi-Fi devices default their settings to the minimum-allowable frame spacing because, again, that benefits throughput. (Go pull up the web page on your Wi-Fi access point and dig into the low-level settings. If you see any IFS parameters, I bet they’re set to basically zero.)
So the reason I think you need “licensing” is to enforce effective power/time limits on the precious spectrum resource. Maybe not the “Google-pays-$1B” kind of licensed–perhaps there’s some middle ground, but it would probably involve the FCC picking a winner in the technology game, and that makes me cringe.
Now, the Libertarian in me says that if you unlicensed the whole damned spectrum, from kHz to visible light, that might allow UWB modems and new-fangled soft radios to render the whole throughput issue moot. But so long as we’re talking about little 100-MHz blocks of spectrum here and there, it just doesn’t seem to work out. The resource is too constrained to enable The Vision.
> The reason Iâ€™m asking is because if more spectrum in different places (e.g. with better non-line-of-sight propagation) can solve the bottleneck, then the problem is subject to political attack by a big player like Google.
So, I’m self-aware enough to realize that, just because we couldn’t solve the problems, doesn’t mean someone else won’t. I just think it’s a much harder problem than most “armchair hackers” realize, and at the heart of the problem is spectrum.
And as it turns out, I just started a new job at Google a couple of months ago. It’s a big company, but if I can find the right ears to bend, I’m on it. :)
# jsk Says:
> Do we eliminate free, broadcast TV (all broadcast TV is now UHF) to feed the desire for
> Internet access wherever we happen to be standing?
“Considering the Internet actually achieves what TV was originally envisioned and intended for all, Iâ€™d answer â€˜yesâ€™ to that.”
I’m not sure what you mean by “originally envisioned and intended for all”, unless you mean envisioned in such a way that everyone has to pay a monthly fee to watch TV. If that’s what you mean then there is no way a mesh network of the type being discussed will take shape. Geezers who never leave their home and people who refuse to pay for cable (and often broadband Internet access) would rise up and revolt. You can also take your idea to the extreme and snatch all the bandwidth from AM and FM broadcasters (including ham and CB folks), as well. But forcing people to pay a monthly fee for not only phone and internet, but to view TV or listen to the radio is just nuts. The whole point of the broadcast spectrum is so people *don’t* have to pay to get access to news, information and entertainment. There is also the anonymity factor. You watch broadcast TV or listen to the radio and no one logs, or even knows what you are doing. Force everyone to get this from the Internet via a mesh network and all of a sudden what you watch gets logged and studied by people who want all up in your business so they can advertise and tailor marketing to you personally. Not only does it give me the creeps to think that some entry level doofus fresh out of college has been tasked to personally induce me to increase my spending by X%, but I consider my viewing, listening and Internet surfing behaviors to be proprietary. If people want to observe and analyze them they can pay me cash. A whole lot of cash. I guess if you are one of those utopians who believe Internet access should be free for everyone the cost argument carries little weight. But the idea that Internet access should be free to everyone is nuts on its face.
> Iâ€™m not sure what you mean by â€œoriginally envisioned and intended for allâ€
Nix the ‘all,’ that’s a typo on my part. Television was originally hoped to be a decentralized communications tool, but turned into something else.
# jsk Says:
“Nix the â€˜all,â€™ thatâ€™s a typo on my part. Television was originally hoped to be a decentralized communications tool, but turned into something else.”
How is it not a decentralized communications tool? Programming producers are decentralized all over the world. Ditto for broadcasting networks. Broadcasting stations are decentralized and spread all over towns, even though all the transmitters might be located on the same high point. Television receivers are decentralized in every home. There is programming of all sorts, good and bad being broadcast to those homes. I think you can still get the equivalent of the Sony Watchman so you can watch TV as you move about. And it all uses the ultimate mesh network: High power broadcasting that penetrates virtually all environments. Where has broadcast, i.e. free television fallen short?
> How is it not a decentralized communications tool?
Sorry, rephrase: decentralized, two-way communications. If you only want to consume, it’s fine. But, frankly, I’ll take the monthly cost of the Internet that allows me to potentially reach and communicate with millions and millions of people over the staggering cost of licensing and power to effectively broadcast via televsion.
# jsk Says:
“Sorry, rephrase: decentralized, two-way communications. If you only want to consume, itâ€™s fine. But, frankly, Iâ€™ll take the monthly cost of the Internet that allows me to potentially reach and communicate with millions and millions of people over the staggering cost of licensing and power to effectively broadcast via televsion.”
The beauty of living in the 21st century, especially in the United States is that things like this aren’t an either/or proposition. We can choose how we consume media. If I’m trying to reach people I choose the best vehicle for that purpose. But when I’m watching Lost or Stargate or some show it’s not a two-way behavior. How would it be different watching those shows if the utopian mesh network was in place, other than I would have to pay to see something that was previously free? You talk about the “staggering cost of licensing and power”, but those costs are paid for by the people who choose to invest in those businesses, and recovered through advertising. And they make a profit, otherwise they wouldn’t still be in business. The whole mesh idea can be disposed of with simple technology. It just takes time. When radio first started each station took up a lot more bandwidth than they do now. Technology improved the ability to broadcast in narrower ranges. That’s why you can have more radio stations, closer together now. Same for digital TV. When the Internet first went public there was no way to play interactive online games unless you loved Microsoft Hearts of some similar minor game. Technology and greater efficiency has made it possible to play all sorts of online games now. The spectrum didn’t get bigger. We got more efficient. The same will happen with portability for mobile devices. You don’t have to take more spectrum to make it work. You just have to wait.
I have Linksys WRT54G, but I can see my AIFSN parameters because I run the DD-WRT firmware on it. You’re sort-of right in my case. DD-WRT uses different AIFSN values for different traffic types through the use of WMM QoS shaping; the values are definitely closer to 0 for some types of traffic. Here is a screenshot of my current values, which I believe are the defaults. I’ll bet the default firmware uses a different AIFSN value.
The traditional solution to this problem is to what Seattle Wireless does: have a bunch of volunteer monkeys to configure and tune everything. Obviously this won’t work on a much larger scale, however.
> DD-WRT uses different AIFSN values for different traffic types through the use of WMM QoS shaping;
Neat! Leave it to the open implementation to do it “correctly.” :) I guess they don’t have a marketing dept. sitting on their shoulders screaming “make it go faster!!!”
> The traditional solution to this problem is to what Seattle Wireless does: have a bunch of volunteer monkeys to configure and tune everything. Obviously this wonâ€™t work on a much larger scale, however.
I guess my point is, in a diverse ecosystem of end-consumer devices, where manufacturers compete in simple terms consumers understand, range, reliability, and throughput drive the features and design decisions. And that means that devices tend toward poor spectral efficiency in unlicensed applications–consumers wouldn’t recognize spectral efficiency if it hit them on the head, and they have no incentive to care for anything but their own packets. A cohesive group deploying a purpose-built network can do it properly (we had several large-scale examples), but the sort of ad-hoc organic collection of consumer devices that the original article envisions isn’t likely to play so nice or, in the end, work so well.
Fascinating comments thread!
I’ve worked in this field too, on experimental roaming mesh-networks deployed in much more constrained circumstances than are being discussed here, and I recognise the technical issues that have been described (power usage at key nodes; latency; coverage). I agree the spectrum issues of such a wide-scale deployment would be insurmountable by the market; regulation would be required. I hope for an eventual move to UWB and adaptive software radios however I think there are too many commercial interests vested in the spectrum-allocation arena for this to be possible in the short-term.
What I haven’t seen yet is a discussion on what happens to the carriers’ charging models and revenue streams if groups of devices can avoid using their fixed infrastructure to communicate. Do devices become more expensive as carrier subsidies reduce? Do charges for backhaul network access increase in proportion to the amount of peer-to-peer traffic? Or would the peer-to-peer capability instead enable communication only in niche circumstances while leaving the fixed infrastucture as the default and majority transport mechanism?
All this stuff about licensing bandwidth would be obsolete if we could get the old guard out of the way, and implement technologies like ultrawideband without ridiculously-low power limits. Way more bandwidth, way less power usage, and not a lot of concern about interference. (Google “open spectrum.”)
If we still have concerns about people eating up our phone batteries unfairly, we could implement some form of digital currency micropayments.
A couple of quick thoughts:
For battery power, have an option which only enables mesh access if the device is currently plugged in. A lot of people charge their phones in the car, at home, etc.
For emergency use, there is a need for communications, not necessarily voice communications. If you can merely route email/text messages reliably and without too much latency (tens of seconds), you can readily communicate with your family/friends to make sure that everybody is okay, or not. We can grow better uses/efficiency overtop of that as time goes on.
> For battery power, have an option which only enables mesh access if the device is currently plugged in. A lot of people charge their phones in the car, at home, etc.
I don’t think I’m giving away any technical secrets to say that there’s actually a lot of cool things you can do in routing protocols if everybody “plays fair.” You can use routing metrics to bias toward/against certain paths/nodes, if those nodes are being honest about their capabilities (power, connectedness, backhaul, etc.), and being honest about their QoS demands. For example, you don’t route through a battery-powered mobile device unless your traffic is emergency-class, like a 911 call. Only problem is, in fully unlicensed systems, there’s nothing stopping you from calling your pr0n downloads an “emergency.” Again, this goes back to “playing fair.” Perhaps there’s some social/market solution to this problem. I’m a big fan of the micropayments idea–I’ve tossed that one around many times–but it’s not clear to me how to keep secondary “undernets” and other protocols from consuming the spectrum.
Oh, and so long as we’re talking about fanciful ideas like UWB, I should toss in beam-steering antennas. Those could dramatically improve spatial efficiency, but they’re really tough to do for mobile consumer devices. Like I said, someone might be able to solve these problems, but I just don’t think we’re there yet.
I just want to throw in one more quick sentence here before I head off to bed, seeing how the topic has veered:
There are a *lot* of not- and not-very-well documented assumptions, both societal and technical, that are baked into our present systems of terrestrial broadcast radio and television, and as we evolve those, we’d better damned well give some pretty high-level thought to what they are.
A good example of what I mean is the assumptions we’re finding break as we slowly transition from a central-battery analog POTS telephony system to one with not only distributed intelligence, but distributed power.
Anyone who doesn’t think we really *need* the positive attributes provided by these historical designs has not been in a *big enough* disaster. (Think: Katrina, and if you need more on that, see here:
As you can read I’m also a big fan of the adhoc net idea. Even though we can see a clear goal, it’s the path forward that we’re caught up in. Your strategy (a backup in case of emergency), is novel and it may have a chance.
All this network needs is two folks and a smart phone app to begin.
As an incentive to allow use of a user’s hardware, perhaps use a digital currency like Bitcoin. CPU power used to route traffic also generates DC credits.
Huh. I thought of this months ago. I just figured that (like most of my best ideas) everyone else had thought of it before. *facepalm*
[quote]By pitching it as a backup in case a power outage or disaster takes out cellphone towers, Dr. Gardner makes it an easy political sell and hard for the carriers to oppose. Regulators might even mandate it as a way to solve the 911 problem![/quote]
Right. Because the people running the telecom industries are f’ing morons and won’t see through this.
Telecom providers, ESPECIALLY cell phone companies are f’ing dirtbags. ALL OF THEM. They are not stupid, they are evil.
I wish something like this would work, I really do–mostly for anonymity and anti-censorship reasons. However I suspect that you’d have a better chance skipping the cellphone companies and doing it just as a 802.11 pocket sized device that had the ability to talk to multiple access points at once, and would auto-connect to access point or ad-hoc network available, then make it “cool” to provision an open, public access point.
This sounds suspiciously like the network from Cory Doctorow’s “Someone Comes to Town, Someone Leaves Town.” But with the ideas from the One Laptop Per Child project thrown in. Awesome.
@JamesFromPittsburgh: yea, Stross’ Halting State was my first thought too. Great book.
This is indirectly aimed at Casey, because for all I know we met way back when I was in Florida and I’ve spent the last 10 years basically working on the same stuff, and would love to talk about where I’m at with what I’ve been working with…
0) I worked on the mesh potato project for a while, gave up for a variety of
reasons. I’d have argued that the design team for the hardware have moved
to Timor before even starting the spec, as I was living in Nicaragua at the
Still, the project shows potential. David rowe had an excellent series of blog entries on the first deployment of the mesh potato:
(10 blog entries, this is the last, with an index to the prior)
1) Power for mobile nodes is a serious problem, but there are plenty of other
ways to build a mesh network using fixed nodes – two nanostation M5s eat .14 ah
of power, for example, easily handled by a small solar cell and a battery.
The power problem is going to take a long time to solve, if ever, and can’t be solved inside
the 2.4 ghz spectrum, period.
2) 2.4 ghz is JUNK spectrum, made all the worse by everything else on it, and everybody piling on it even more. Bluetooth doesn’t even work well between your belt and your head nowadays.
3) Everybody is too USA centered. Wake up, people! 4/5ths of the people on the planet don’t have internet yet. There are cheap, and mesh based solutions to many of the problems that the rest of the world faces. (mesh is useful in a certain circumstances, anyway)
4) USA’s IP regime and licensing standards are too hard to cope with. I spent
95% of my time in my last 5 years here working out ways to avoid patents
rather than actually doing engineering.
So in answer to all this, I left the USA, moved to Nicaragua, and began to
develop a greenfield, 5.8 ghz, ipv6 enabled, meshed network, and got it
working a few months back on the nanostation m5’s with openwrt… (as well as a few other
pieces of hardware – the sheevaplug and ubiquity 2HP primarily)
The nanostation m5s cost 84 dollars each. Currently. They have a stated range of 12km. (I have tested 6km under good conditions) They can do 802.11n (300 mbit!) and 802.11a…
Nica has no rules regarding this spectrum and nobody using it in the rural (and mountainous) areas I was targetting. I imagine this is the case throughout much of the non-developed world….
I’d deployed 3 nodes and was working on getting the rest of 11 initial nodes
deployed before the rainy season hit… so I could test thoroughly in the
most terrible conditions possible…
…when the survivor tv show rented my house out from under me. It’s a long story.
I have been travelling the USA with 3 nodes in my backpack ever since, trying
to figure out who to talk to about getting the test deployment done.
I already have access to a 3 towers down there to test with, and another dozen, readily available…
For my network map, there’s a google earth map here:
http://www.teklibre.com/~d/b4barrios10.kml and zoom in below San Juan Del Sur,
Nicaragua. (it puts you in the center of the planet for some reason, to start)
For the most recent generation (sorry, untested in the field) of the code, see http://opkg.teklibre.com/
The Babel protocol is neat, it does both ipv4 and ipv6 cleanly…
So is ahcp…
I’m in Redwood City, California next week….
It appears that this is now happening….
>It appears that this is now happening….
Read the article, saw the word “proprietary”, and immediately lost interest.