NYC Mesh is trying to get around the big ISPs one node at a time | Hacker News Reader

NYC Mesh is trying to get around the big ISPs one node at a time

technical.ly

127 pointswinstonford46 comments

There is an interesting symbiosis developing between NYC Mesh and my startup Canopy: the data mesh wants power, the power mesh wants data. Both want the other to be more resilient and less expensive.

In the pic at the top of the article, taken on my rooftop, you can see a white, round thing. This is a ubiquiti nanobeam wifi radio. Just below it you can see a solar panel. It belongs to an off-grid solar power unit called Ra http://canopyrising.com/#ra. The nanobeam is now bolted directly onto the solar power unit Ra enabling the node to support the local data mesh during a blackout.

On the roof and inside Ra, a beaglebone black runs the http://sunpress.co network of websites directly on off-grid solar power.

Ra also delivers solar power for lighting and usb charging in the form of the Canopy Cuba table http://canopyrising.com/#cuba , a bedside table that stores days worth of power and is found in the http://tremolino.co guest rooms 1 floor down.

Canopy aims to create a marketplace where neighbors who invest in the products to generate solar power trade with those who invest in capacity to store it.

Unlike traditional solar and storage, which is limited to property owners as it requires financing, design, permitting, and installation; Canopy is accessible renters and owners as it is low cost and requires no financing, no design, no permitting, and no install.

How does the cost and resiliency compare to purchasing energy from the grid?

For reference, I buy 100% wind source power from Xcell @ $0.14 / kWh. Reliability is generally very good, but to compare apples to apples it would cost ~$500 to purchase a backup generator [1] for the 1-2 hours of outages we have every year.

[1]: a battery system would fulfill a similar role, but I'm not sure of the cost -- and for something that I would essentially never use, it may actually be better for the environment to keep the generator.

Great question @oconnore. 2 answers.

1. The target price for Ra http://canopyrising.com/#ra is $600-800, just under what you would pay for a 1kw emergency generator. Initial market is apartment renters in flat-roof, multi-floor urban buildings, who can not run generators indoors b/c of fumes, etc.

2. Battery capacity built into the Cuba table http://canopyrising.com/#cuba allows power arbitrage, buying power at the lowest price during off-peak hours to charge up, then drawing from batteries during peak pricing on the grid. In NYC, Conedison sells power for 7-19 cents during the day, but from midnight until 8a, it costs just over a penny. Voluntary time of use programs are common across the country, but few products exist to leverage them. http://www.coned.com/customercentral/energyresvoluntary.asp

Hi. I do research on Guifi.net (part of www.clommunity-project.eu), referenced in the article. Guifi.net has 30k+ nodes, and provides Internet access. We have also built decentralized services specifically for these type of "Community Networks" - including a decentralized search services, and a Video-on-Demand service (www.decentrify.io - to be released this month). Happy to answer questions.

This is very interesting, but I wonder: how do you guys handle abuses? Especially when it comes to sharing the home connection to the 'outer' Internet with strangers. Do you keep logs? If so, how do you handle the privacy part?

I think in part this is a problem that also Tor relay owners have (minus the log keeping obviously).

Sharing the home connection outside is encouraged, in fact. There are logs, of course. Individual nodes can be kicked out if needed, as all routing is BGP based.

Does each node hold a full BGP routing table? That could be a problem as the number of nodes increases

Nope. There are are domains. What's also interesting is that GPS coords are used extensively for joining domains and accessing services nearby. We have our own UNIX distribution now for cloud services. It's Called 'cloudy'

Totally awesome, thanks. Could you break down costs (ideally in dollars) for getting a network up and running. To clarify, how much would infrastructure cost initially for the setup and also to the end user. How much did it cost to connect those 30k nodes, how many people are served and what kind of speed and bandwith are you seeing?

Sorry to lob a ton of ?s at you but this is SUPER INTERESTING.

Thanks.

Ordinary users pay 59 euros (65 dollars) for a wireless access point. Typically they pay somebody 20 euros to install the directed antenna outside the building. For directed wifi connections, you can get up to 20-50 Mb/s or single digit Mb/s. Some links are very long - up to 10 kms. Obviously, over many hops it degrades, but most traffic goes to the Internet gateway right now - which is never too far. There are also users around the network with outside access who offer their machines as Internet proxies. Many of these are used, so a lot of traffic terminates inside Guifi.net at those points as well.

The cost was born by people who wanted cheap Internet. 65 bucks for free Internet.

Wow totally fascinating. Thanks for answering. Would be great to see this take off in NY.

Does the mesh network have enough bandwidth to link distant cities, or do you just use the mesh for the 'last mile'?

For example, if there were mesh networks in New York and San Francisco (which are a few thousand miles apart) I would have thought the link would get congested quickly, if you could establish it at all.

If you use something else for backbone, what do you use and how many users does it support / how many users do you need to pay for it?

It's a mix of 802.11 directed antenna connections and now fibre linking up large areas with downtown Barcelona where the connection to the outside Internet is located. There are large valleys with poor 802.11 connections to Barcelona, and now many of them are getting fibre. Volunteers have learnt how to splice fibre and negotiate where to lay fibre :)

For revenue - there are about 20 ISPs who pay for QoS. Their contributions fund free internet for the other users. They get a guaranteed large share of the available outbound b/w.

Interesting decision to not have it connected to the full Internet. That makes it essentially a local, social, network like the old BBS scene. You use the mesh to talk to people, not so much for streaming media. That will save a lot of bandwidth. It also keeps it enthusiast-only and reduces the need to work out how to handle abuse.

I hope these aren't enthusiast-only. The idea of a 311 type service for a community with craigslist type services, job postings, city info and social aspects would be great for low income people. You could put all of wikipedia on it and partner with Khan Academy to give free access to learning to every kid in the city.

Hi I'm Brian at NYC Mesh. Yes it works as a local social network with our ".mesh" addresses, but most nodes are also internet gateways so they act as normal wifi access points. https://nycmesh.net/faq/

I bet they can eventually build a co-op to pay for connectivity from the Mesh to the CO. Basically a community paid for high bandwidth connection out of the mesh to the internet. Which basically turn the mesh into a last mile network.

In Germany, there is http://freifunk.net/en/ (the English wepage is a bit sparse) which has the capabilities to do mesh networking where nodes are dense enough. Since this is rarely the case, it's a great way for people to share their internet connections with the world without taking on any liability. All Freifunk data is routed through their VPN, and they are registered as a provider, so they're not liable under Störerhaftung.

One of the strategy NYCMesh took for where line of sight was not available was to tunnel through the Internet to mesh. This would enable anyone to join the mesh without line of sight requirement, and once more and more people join and the density increases, they can get their dependency on Internet Tunnels away, by connecting/hopping through line-of-sight nodes.

Warning, only Freifunk Berlin is an ISP, all other Freifunk communities tunnel through a VPN to Switzerland.

Source: I hang around in the #ffki channel on HackInt for the Freifunk Kiel community quite often.

Reminds me of the old days of KA9Q and pushing IP over packet radio. Tech has come such a long way this might actually be a real alternative. That said bandwidth management will become a real issue quickly since it's hard for the general public to understand the shared nature of a mesh.

Mesh networks are for the birds. For those who want to enable codependent telecoms to not show moral fiber they are appealing, but one side way to make a radio network slow and unreliable is to retransmit and retransmit and retransmit... If I wanted to have bad internet I can call up my telco and pay an arm and a leg for DSL.

If only there was an entire class of protocols developed over the last 30 years to negotiate and route network traffic efficiently.

These work well for wired networks, but not for ad hoc wireless networks. Mesh networks are on of the many areas of CS that produces endless conference proceedings but no working systems because the DOD has the hots for it.

There a many huge mesh networks all over the world, just not so many in this country. Guifi in Spain has 29,208 active nodes so there is no argument that mesh doesn't scale. https://guifi.net/en/node/38392

I can't find much in the way of specific technical information in English, but from what I can tell Guifi has are tiers of nodes, ordinary nodes and supernodes, where nodes connect only to supernodes and the supernodes have fixed, manually configured directional radio links to other supernodes. It's a lot easier to get thirty thousand nodes on a mesh network if the vast majority of them aren't actually taking part in the meshing and the ones that are use fixed infrastructure links. It basically avoids all the inconvenient "mesh" part.

Awww, I love this idea... I am sitting in a cafe in Brooklyn reading the article. I checked the available networks and found an "nycmesh" ssid. Went to ev.mesh and got the "this is not the internet"! So cool, but I tried to create an account but got "Error message: can't send email", and I can't log in.

Still, I love the idea though - feels like a secret internet club!

Hey, do you happen to be in Dumbo?

Yup! Spooky! Internet on NYCMesh is much faster than WelcomeBRC ;)

you, sir, are on my node. enjoy!

Oops, that is my server! I'll fix it today. What café are you in? We don't even know about this access point. Try chat.mesh

Are you at Brooklyn Roasting Co?

You could support Netflix by adding a Bitcoin-based system of transit settlements. This could provide the capital to really scale up, without any central control or administration. I

Of course the cooperative part of the network remains free. A payment system just makes expensive infrastructure possible as part of the system, like fiber under the oceans. A decentralized payment system could allow massive scale by allowing investors to finance and make money off network segments, repeaters, etc. Real competition forces them to compete on pricing and quality. One can imagine a smart routing algorithm that takes cost and quality into account.

I mention this because I'm a skeptic about many proposed applications for Bitcoin, but infrastructure with no central control is a great use case. Bitcoin needs some solid use cases before it can really take off.

Edit: any comments along with the downvotes? I was hoping for a conversation. I know its crazy but I'm speculating about creating an entire new internet.

100% agree. Of all the potential use cases for Bitcoin, decentralized protocol-level payments (for network bandwidth, torrent seeding etc) are one of the most compelling.

I've seen this being done for torrents already, but would love to see someone try it out for negotiating bandwidth for wifi/mesh networks.

Maybe something more along the branch of Namecoin, but optimized for micropayments to pay bandwidth providers daily rates?

Thinking long term, setting some type of expiration system after coins are mined, by a set decomposition rate into the protocol to solve both the hording/anti-economy issue inherent to bitcoin's hard limit on coin supply. Coins expiring might potentially help with ledger bloat? Avoiding potential abuse of "juggling" the coins by transferring them between several wallets to avoid the decomposition rate would be the first issue that comes to mind.

Really cool idea. Does anyone know about the legality of sharing internet through meshes as referenced in the article, "RHI pays its internet subscription to Brooklyn Fiber, and redistributes this coverage for free to a dozen parts of Red Hook." ? Or whether ISPs would be able to shutdown mesh networks for sharing internet subscription like this?

How different is this from backbone peering?

It's certainly "legal" and RHI would only need worry about whether it's permitted as a part of their service. Presumably it is permitted, but certainly not all providers would allow it. Baked into the cost of the service no doubt.

Brooklyn Fiber is a commercial provider, so buying access from someone like them is basically how an ISP becomes an ISP.

If you were sharing your Cable or DSL, you might run into issues with your contract.

In both cases you would have to be prepared to handle abuse complaints, or your IP space would likely just get blocked from large parts of the internet, or at least major services.

For instance, the administrator of an office network may receive DMCA complaints about employees downloading movies or somesuch. If these continue, at some point your internet provider will terminate your connection. The same would be a concern for anonymous internet sharing.

We tried something similar in many cities around Ireland in early 2000's when the first 802.11b antennas were coming out - https://web.archive.org/web/20030420231414/http://www.dublin...

What would it take to set this up in a new city? It would be awesome to have this in SF as well.

This might interest you http://www.broadband-hamnet.org/ it's a mesh network using modified wifi access points for setting up a very high bandwidth mesh network. Quite a few people in my area (Pleasanton, Livermore) are using it. I'm sure there are people in the SF area. This website would be a good starting point for finding them. Also there are some of the worlds best ham radio clubs down in San Fran and the South Bay the towuld be a great place to start in finding people that want to get serious about mesh networks. Another related field that is fun is microwave community cations, and I'm pretty sure there are some active groups in your area for that.

There's a commercial version of this in parts of Boston: http://www.netblazr.com/

Yeah, but their coverage is poor. Comcast has Boston completely strangled.

Boston is strangled in many ways. Miraculously, they evaded the Olympics.

There's a small group of us in Boston working on a free meshnet as well. We're hanging out on #bostonmeshnet on Freenode and there is a mailing list here:

https://lists.projectmesh.net/cgi-bin/mailman/listinfo/bosto...

Reply on news.ycombinator.com