Hacker News new | past | comments | ask | show | jobs | submit login
Hyperlocal Mesh Networks (avc.com)
136 points by andreyf on Aug 24, 2014 | hide | past | web | favorite | 31 comments

They use mostly Ubiquiti radios, antennas, routers and wifi access points. Ubiquiti Networks is quietly revolutionizing the internet around the world. They may actually turn the internet into the all-encompassing democratic communication tool it was supposed to be.

They also use the commotion (https://commotionwireless.net/), an open source software project that runs on commercially available wifi access points and routers and creates the mesh network.

They also use tidepools (http://tidepools.co), a platform designed alongside the Red Hook mesh, for local apps on the network (stop & frisk, maps etc.)

Case study: http://www.scribd.com/doc/147627564/Red-Hook-Initiative-Tide...

Over the past year or so I have watched from afar as wireless tech is highlighted sporadically. I have to wonder if it's only comes to my attention when someone makes a marketing push. I'm not dismissing it as pure hype, as it's definitely captivating from just an engineering viewpoint. Whether it's hype or not, this type of cheaper decentralized wireless networking seems very cool and possibly ripe for motivating new business models. There are a few of companies (and their founders) making noise and at least one interesting acquisition not too long ago:

+ Ubiquiti (founded by Robert J. Pera, hn might appreciate his entrepreneurial background, look him up) + Artemis (founded by Steve Perlman, another hacker/entrepreneur ) + Meraki (aquired by Cisco in late 2012, seems like this one did actual mesh networking in MIT and SF)

It seems like there is a lot of noise around these wireless technologies, what is holding back adoption?

I spent most of 2001-2008 involved with community and then commercial mesh projects. The ideas are attractive but there are some technical realities that make deployment unlikely unless there is no alternative. In most places most people want to connect to the regular web and wires (or fiber maybe with point-to-point wireless for the last hop) are faster and more reliable.

Existing 802.11 ad-hoc is half duplex & CSMA, if you build a multihop mesh with that on omnidirectional nodes you get greatly diminished performance across the network. TDMA is not really better, the whole thing degenerates into a distributed graph coloring problem. You can have a lot of success building a network of point-to-point links but again you only really see people getting excited because they need non-wired infrastructure or because they like the technology. The Ubiquiti gear is really good for this.

Meanwhile you might wonder about doing this on small battery powered devices. It's very difficult to build a mostly-asleep power management scheme if a device needs to be awake to relay packets.

I still think there's promise in the area though and do follow what other people are doing.

If you want fast access to the general internet, mesh networks are never going to be particularly competitive with telcos. This is what most people want, as well.

Community networking over community networks could be a thing, but bootstrapping is hard, and it can be hard to see the point. "Buy this thing and set it up and you can talk to people within shouting distance with your pc/phone" isn't super exciting.

It seems like the problem is that it's easier to make the normal internet produce the same behavior than to actually make a mesh network fast and reliable. If all your neighbors have normal internet with NAT-PMP supporting gateways then you can create a "community network" entirely in software.

Maybe the problem is they're doing it from the wrong angle. Don't create a mesh network independent of the internet, create a mesh network integrated with the internet. Then use whichever interface has the best performing route and use them to back each other. So if your wireless goes down you can still use Comcast and if your Comcast goes down then you can seamlessly use your neighbor's Verizon.

Ubiquiti makes it easy.

I set up a point-to-point link a few weeks ago using two of their cheaper antennas - it took minutes to configure the encrypted, secure connection - it "just works".

I'd love to join a mesh network (or a p2p node network) where I live, but there doesn't seem to be one available.

In Melbourne, Australia there's a rather large wireless network that's existed for years - but I can't find similar networks in any other of the country's large (or even smaller) cities.

There have been a number of attempts to get one going in Canberra. If anyone knows anyone who wants to try again (in Canberra or elsewhere), please get in touch. I may not be able to help directly but I can probably put you in touch with people who can.

Contact details are in my profile.

Yes - a project I worked on used exactly that - Ubiquiti and Commotion (milemesh.com)

I really enjoyed this HOPE X talk on this subject: https://www.youtube.com/watch?v=cEo4uLAJ32U

http://Guifi.net in Spain has 20K+ nodes in their mesh network, http://rising.globalvoicesonline.org/blog/2013/12/11/guifi-n...

"guifi.net [es] is a user-owned, open and neutral network in which a growing community of volunteers can connect their computers to form a sort of intranet and, at the same time, share an Internet connection. The non-profit network is free, minus any individual costs for networking equipment, and anyone is allowed to join and use it how they want."

More details: https://lists.thefnf.org/pipermail/discuss/2014-March/001672...

Edit: may be coming to the US, https://thefnf.org

Every time I see these posts pop up about mesh networking I'm more and more intrigued of the idea of a local neighbourhood network. I live on a small island of a few 1000 people. It would be brilliant to extend the sense of community with basic services like a marketplace, events etc. It kind of reminds me of using CB radios in the UK in the early 90s, communicating with local people only.

Is there any defacto hardware out there that one can buy to start the mesh, without a huge setup, that I could then set up for a couple of neighbours without hassle who then automatically join the mesh? For something like this to take off it'll need to be as simple as picking up a router off the shelf and reading a few lines of a manual.

What are the best resources to learn more about mesh networking? I've been fascinated with it and trying to soak up as much information as possible.

I'm trying to understand what advances there have been in reduction of speed over multi hop networks. That's the biggest issue I see, is that people want high speeds, but mesh networks at a large scale might make that difficult. Multi-frequency radios can only go so far.

Wireless Networking in the Developing World was a good read: http://www.wndw.net/

The Free Network Foundation has aggregated a lot of practical info: https://commons.thefnf.org/index.php/Portal:Education

A link from the FNF on multi-hop optimization: http://researchrepository.murdoch.edu.au/3982/1/Comparison_o...)

The motherboard documentary about the FNF is a must watch for any mesh network enthusiast: https://www.youtube.com/watch?v=Fx93WJPCCGs

So if mesh networks take off, and replace mobile internet for many people, are we going to see an increase in satellite and "traditional" mobile networking (3G, 4G, etc) prices?

How is that going to affect folks outside major metropolitan areas, since their coverage is effectively subsidised by the volume of city users?

Likely a decrease, 3G is cheap, mesh networks would merely highlight the bad deal most telcos provide.

To be fair, you'll see increases even if this doesn't happen.

More internet connected local mesh networks like this please! We've got something similar here in Austria as well - http://www.funkfeuer.at has been growing constantly for more than 10 years.

Voice of America article: http://www.voanews.com/content/austria-programmers-build-fre...

"A group of computer programmers and hackers in Austria is creating a low-cost way of spreading Internet access across communities. FunkFeuer which means network fire in German, uses everyday technology to create a wireless network, called a mesh, that can transmit data from person to person, without involving companies or governments."

Wasn’t the problem with mesh networks that the amount of meta data needed eventually becomes too big?

yah, this is really curious too! i thought the amount of 'mesh' traffic goes up with the number of mesh nodes..

It's not a traffic [bandwidth] problem, really. You have two basic kinds of wireless mesh networks: infrastructure mesh, and ad-hoc mesh. With the former, you can design the network so that each directly-connected node has a dedicated channel and you preserve full bandwidth across the spectrum. You also have only a few gateway nodes so your network updates are very few. With the latter you might be using one channel in half duplex to communicate with whatever nodes are closest to you and propagate joins/parts throughout the network. That will have less bandwidth and be slower to communicate as you add nodes, partly due to number of additional hops.

If the number of nodes who deal with routing increases, that's more nodes that need to be informed of each join/part, so technically it takes longer to update the network. But if the joins/parts are few and the signal is strong, this is a rare event. More nodes can make the whole system faster, or it can make the whole system slower. It depends on the implementation. But there's not a constant flow of mesh routing data that multiplies with nodes; that bandwidth used is tiny.

Are the networks you talk about tamper-proof? I vaguely remeber that any attempts at making ad-hoc mesh network temper-proof didn’t succeed.

I've never heard of a tamper-proof network. If you've got a network, I can tamper with it.

Ok, not proof but reasonably resistant.

Can these mesh networks be easily joined with other mesh networks to build a global mesh?

Would pCells work in public spectrum?

> Will it work for WiFi as well? It's protocol agnostic, so it could work in unlicensed spectrum as well. The issue is that you don't have complete control over all the other transmitters, so you can't coordinate them.


I think mesh networking is a hugely important tech.

From the article:

"Most mesh networks are connected to the public Internet, but if that connection goes down, the local mesh continues to work. In Red Hook that means that you could make voice calls (over IP) from your housing project to the local hardware store to see if its open. Or you could email a friend who lives in the neighborhood."

I don't know VOIP very well so this may certainly be possible, but I very much doubt that without an external internet connection you'd be able to email anyone.

That could easily be hyperbole on the author's part but you don't need internet if Red Hook residents are running their own mail servers. Since they could also possibly be running their own VOIP servers it isn't terribly far fetched.

Multiple non-internet based servers can easily talk to each other as long as they have a map of how to do so. RHI could be running their own DNS in which case each set of sendmail / qmail / whatever servers could still see each other. And if they really want to go old-school they could use UUCP or NNTP to shift around e-mail and news respectively between sometimes connected nodes. Many of the "intermittent internet connection" problems have been solved for a while.

Email works on a single server, it's pretty easy to get an email system working that doesn't involve the Internet.

It actually works across multiple servers as well and e-mail exchange predates the "always on" internet everyone's used to. It is feasible RHI is running their own DNS and if they really want to go old school could always support UUCP.

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact