Project Meshnet: Hardware for cjdns to enable new, censorship free Internet

[Tirapon]

From what I understand the point of a mesh network is the decentralization - if you have many users, taking down a few nodes won't help because you can just route around it - you would have to take down a significant percentage of nodes to disrupt the network. In terms of disrupting your signal, my knowledge isn't good enough to tell you how feasible that would be.

But in a local area such as a small town how many nodes are there going to be? Surely with only one or two possible routes it would be easy to take down. Very few in one small area are likely to be activists.

That's the difficulty of developing a mesh network - getting enough people involved to make it work. If you could get the majority of the population using it, it would be incredibly resilient. I guess like most things, offering some sort of financial incentive would get things rolling - just look at the growth in hash rate of the bitcoin network...

[1714795199]

1714795199

[1714795199]

1714795199

[1714795199]

1714795199

[ElectricMucus]

From what I understand the point of a mesh network is the decentralization - if you have many users, taking down a few nodes won't help because you can just route around it - you would have to take down a significant percentage of nodes to disrupt the network. In terms of disrupting your signal, my knowledge isn't good enough to tell you how feasible that would be.

But in a local area such as a small town how many nodes are there going to be? Surely with only one or two possible routes it would be easy to take down. Very few in one small area are likely to be activists.

That's why modulated light communication is so great.
With radio signals in wifi and co it's possible to disrupt it with jamming devices, less so with high gain antennas but it's still feasible.
Ronja on the other hand can only be disrupted at the node itself or in a direct path. The only thing that really affects it is dense mist, and to my knowledge there is no technology which could generate that amount since it is built to handle natural conditions you might encounter.
It has several advantages over wifi which would be too extensive to mention here, the best of which is that it can work as a backbone for a traditional non-directed wifi network because there is no interference.
And 10mbit/s is only the speed of a naive implementation, in theory you can reach almost fiber speeds.

[bytemaster]

The key to getting a mesh network to work is the decentralized hosting of data.   Minimize the client/server communication and turn the internet into one giant torrent.   This would allow the network to scale to any demand.

Prior to working on BitShares... I had the following ideas:
https://github.com/bytemaster/tornet

To prevent Sybil attacks and encourage long term establishment of node identities, each node is ranked by hashing its public key with a nonce, the lower the resulting hash the higher the rank. It requires significant time/cpu resources to generate a high-ranking identity.

Economic Incentive

Price negotiation is 'expensive', especially if you must negotiate with 1000's of peers and the market prices change frequently. As a general rule, most users can pay 'in-kind' by uploading as much or more than they download; however, some nodes have higher demand or are unable to serve files (because they are behind a firewall). Other services are 'asymmetric' such as routing, tunneling, etc and therefor you require more resources from a specific node than that node requires of you. Some users simply want higher speeds and lower latencies.

Therefore, each user simply picks an amount to donate to the network. This donation is then divided among all peers proportional to the amount of service they have provided.  The result of donations is higher ranking in the priority queue and therefore faster network speed and lower latencies.

Each peer extends credit to 'new peers' proportional to their rank and past payment history. Because credit is tied to 'rank' and rank requires cpu time / money to acquire it is not profitable to constantly create new IDs as new IDs have the lowest priority.

An ID that doesn't contribute resources or BTC and uses up its credit will eventually find the network unusable.

Distributed File System
Content is divided into 1 MB chunks that must meet a statistical measure of randomness.  The chunks are stored in a distributed hash table (Kad).  
Each node has a financial incentive to extract the most value out of its limited bandwidth and disk space. A node with limited bandwidth but large disk storage would want to store 'rarely' accessed files that the users looking for them will pay a 'premium' for. A node with unlimited bandwidth, but limited disk space will want to store files in high-demand until the demand for their smaller set of files equals the available bandwidth.

At the same time that nodes want to optimize profitability, we want to ensure that nodes keep content close to their ID. The closer a chunk is to a node's ID the more value that node realizes for providing it. This means that instead of optimizing on access frequency, bandwidth, and storage each node has more incentive to offer chunks near its ID than far from its ID.

The side effect of the above relationship is that each user is incentivized to find the chunk on a node furthest from the chunk ID. This encourages the use of 'binary kad search' of the network and it is this binary search that allows nodes on the network to estimate the popularity of content and then opportunistically cache content that would be 'profitable' for it to host based upon access frequency and distance from the node ID.

On the other hand, nodes that are willing to pay a 'high price' for low-latency can short-circuit the lookup process and start their query much closer to the target node. This short-circut of the lookup process 'harms' the network by hindering the ability of nodes further away to cache the content. If every node did this then they could DOS the target node. Fortunately, because it is more expensive it naturally self-limiting.

Publishing Content

The cost of publishing content on a node is proportional to opportunity cost of that node giving up a slot in its file cache for your content. After all, nodes are in this for profit so each node can multiply the access frequency for a chunk by the yield of that chunk and the determine the expected revenue per-week. A node must 'bump' this chunk in order to pubish the content.

Furthermore, each node can only publish on nodes of 'lower rank', therefore new users / IDs end up being 'beta-testers' for content and nodes that wish to publish to more reliable nodes must invest in their identity. Market forces will then ensure that 'good' content is kept, and 'bad', unused, or outdated content is dropped. Due to the large number of users with small upload speeds and large harddives, there should be significant incentive for them to speculatively store infrequently accessed chunks.


If you combine these features with a 'mesh network', then you solve many of the bandwidth problems because people can pull from nodes near them and data for popular content will be cached.   No one needs to maintain a server to 'publish their data' once it is popular enough to be self-sustaining.   Everyone gets paid for contributing resources and there is financial incentive to contribute.

[ElectricMucus]

The key to getting a mesh network to work is the decentralized hosting of data.   Minimize the client/server communication and turn the internet into one giant torrent.   This would allow the network to scale to any demand.

Great, but an increasing part of the web is dynamic content which wouldn't really benefit from caching. For bandwidth intensive thing like streaming video it's better to support multicast routing.
Caching should be part of the protocol, if somebody wants it they can use usenet, tor or freenet.

From a usage/cost perspective it is almost always better to increase bandwidth instead of caching.

[bytemaster]

The key to getting a mesh network to work is the decentralized hosting of data.   Minimize the client/server communication and turn the internet into one giant torrent.   This would allow the network to scale to any demand.

Great, but an increasing part of the web is dynamic content which wouldn't really benefit from caching. For bandwidth intensive thing like streaming video it's better to support multicast routing.
Caching should be part of the protocol, if somebody wants it they can use usenet, tor or freenet.

From a usage/cost perspective it is almost always better to increase bandwidth instead of caching.

You will never get a P2P hosted youtube with your approach.

Aside from real-time communication, 99% of the data on the web is static.   The other 1% is basically used to index into static content.  For example, google searches could be done in a 'static' manner if their index were published rather than providing the 'service'. 

99% of this forum is static.
99% of changes are 'patches' to already published data. 

The key is to offload all of the redundant data, have distributed hosting indexed by distributed DNS.   Then allow servers to provide the dynamic data that will reference the static data.    If I have to have my server 'always on' to publish data on the mesh network then that is a single point of failure and I risk being identified because all traffic ultimately routes to my server. 

[ElectricMucus]

The key is to offload all of the redundant data, have distributed hosting indexed by distributed DNS.   Then allow servers to provide the dynamic data that will reference the static data.    If I have to have my server 'always on' to publish data on the mesh network then that is a single point of failure and I risk being identified because all traffic ultimately routes to my server. 

Ok, but why re-invent the wheel? A node operator could run a transparent proxy server if he wishes to offload some traffic.

[BTCLuke]

I don't really think meshnets are achievable as the anarchists would like them.

I wholeheartedly disagree.

This is because you really need public and especially government support to pull something like this off.

So bureaucracy has the power to do things that entrepreneurs can't? Lulz...

These projects are publicly against any form of government and sees the government as the enemy which isn't helpful. I see this as being quite dark (notice the word darknet) and negative.

Negative? Really? Ok then, let me ask you this... Which of these two organizations are more negative:

1. The one that has murdered over 260 million people in the last century, while extorting the rest of us all with threats of violence and jailtime if we don't pay our extorted fees, all the while running global data centers that record every electronic communication that we all make.

or

2. The peace-loving one that would like to communicate freely with each other, and will even pay for every bit of it all by themselves.

?

The only reason you feel darkwebs are negative is because you still believe in the gov's propaganda.

Usually the aims are ridiculous and unachievable such as joining routers together with your neighbor who just isn't interested in such things (!), creating networks with helicopters, secret jet-packs or lasers!

Technology always grows to improve these things. The Spanish meshweb has over 40,000 users all connected over a very large distance, covering thousands of square kilometers right this minute... If they can do that with today's technology, In five years we should be able to do it globally.

If a big project like this is to be successful it has to have something for everyone in society that they can contribute to and feel like it has their best interests at heart. I think we probably have to work together on this however difficult.

No. For something this big to be successful, ALL it needs to do is offer a cheaper alternative to what we've got now.

At some point we'll have residential-ready, open-mesh router boxes with 1km+ range cheaply available for households with as much or more bandwidth than we can get from our dinosaur ISPs... All offering FREE internet once you buy the hardware.

How could the ISPs possibly survive that? Who would want slower tech that isn't connected 100% of the time that comes with lots of government eavesdropping and shutoff switches... That you have to pay monthly fees for? Insane.
 

The wireless nature of mesh networks surely makes them useless from a censorship perspective. All the enemy has to do is disrupt your signal or take down a few nodes.

Clearly you don't know what a mesh network is at all. A good demonstration of Mesh networking is presented here at the One Laptop Per Child features page:

http://laptop.org/en/laptop/hardware/features.shtml
(Click on the "start demo" button on the left to start it up.)

That demo only has about 10 nodes in it though... We're trying to network 7 billion people here, and places like spain and athens already have 10's of thousands meshed now.

At least with the internet you have the added advantage of societies dependence on the network for business, which means you just can't filter out what you like from what you don't.

There are already lots of businesses in spain and athens using only their mesh connection to the internet to do business on. Some of those only serve customers inside the meshed community, while others use it as a gateway to the WWW so they can serve the general internet too.

...But once the mesh goes global, who would possibly want to use the old internet anymore? Barbaric.

[Tirapon]

...These projects are publicly against any form of government and sees the government as the enemy which isn't helpful. I see this as being quite dark (notice the word darknet) and negative...

...If a big project like this is to be successful it has to have something for everyone in society that they can contribute to and feel like it has their best interests at heart. I think we probably have to work together on this however difficult...

A completely free and open internet is probably the most powerful tool against oppression that has ever existed. Hopefully enough of society will recognise this for such a project to succeed.

[UncleBobs]

These Flutter gizmos sound interesting:

No that's not good enough.

For a serious wifi alternative look at http://ronja.twibright.com/

Just interesting, not "ready to go". They only just finished their Kickstarter campaign, who knows what they might be able to do in a year or two?  Looking at your link now, thanks.

Anyway, I know there is plenty of skepticism about mesh networks, on very good technical grounds, but I believe the old "where there's a will there's a way" maxim applies.  There is certainly a growing will.

Here's a good "reply to skeptics" article:
http://thefnf.org/why-wireless-mesh-networks-will-save-from-censorship/

[greBit]

Whilst I find the idea of Meshnets fascinating and I wholeheartedly agree that they are well suited to some niche applications but for everyone else what will it offer?

I mean what is the importance of the connectivity layer? If we ignore the 'cool factor' what benefit would I get from connecting through a meshnet than say using my existing commercial provider?

For subversive applications I can use cryptography, VPNs / Tor / I2P etc. I can also completely emulate the meshnet by creating a P2P overlay over my untrusted connection provided by EvilISPCorp.

For a censorship-free internet I think the focus on DIY-internets is a little misplaced. We should look more into developing better tools that work over our existing infrastructure, that keep us anonymous and free through the mathematics of cryptography, not by putting routers on our rooftops.

[ElectricMucus]

Whilst I find the idea of Meshnets fascinating and I wholeheartedly agree that they are well suited to some niche applications but for everyone else what will it offer?

I mean what is the importance of the connectivity layer? If we ignore the 'cool factor' what benefit would I get from connecting through a meshnet than say using my existing commercial provider?

For subversive applications I can use cryptography, VPNs / Tor / I2P etc. I can also completely emulate the meshnet by creating a P2P overlay over my untrusted connection provided by EvilISPCorp.

For a censorship-free internet I think the focus on DIY-internets is a little misplaced. We should look more into developing better tools that work over our existing infrastructure, that keep us anonymous and free through the mathematics of cryptography, not by putting routers on our rooftops.

Currently it only makes sense for the Geeks as existing mesh networks are often organized as a non-profit association which requires some effort and an admittance fee to pay for outside connectivity and stuff.
Although it doesn't necessarily has to be that way. If the trend holds sooner or later there would be commercial kits anybody could use with some state of the art protocol for network discovery. Pretty much the way satellite vs. cable television works now. Main motivation would be reliability, independence and cost.

However I don't think that existing technologies would be the right candidate for mesh networks, although the majority uses it now. It's prone to several difficulties because wifi wasn't made to be used with high gain antennas in a grid architecture. I would prefer free space optical communications for reliability and performance.
I think if somebody were to commercialize it for the consumer sector this could go a long way. These things could be available for about $200 end market price and would fly off the shelves if there were a protocol which is both easy to use and efficient enough.

[BTCLuke]

but for everyone else what will it offer?

4 things:

1. Free internet. (As in zero-cost) There are tens of thousands of people using their mesh networks in Europe and elsewhere today as their only way to get online; and all they paid for was the hardward costs... No ongoing fees of any kind.

2. Free internet. (As in liberty) Don't like how you ISP is stopping you from downloading torrents or porn or something else they feel is wrong? Sorry, obomba told them to do it. So why should you have to go through the trouble of encrypting things like torrents when you could just do away with ISPs totally?

3. Small interruption avoidance. If a truck or lightning bolt hits into the wrong telephone pole up the road from you, you'll be without internet for hours or days. Not so with a meshed web.

4. Large interruption avoidance.  As little as you personally think this is a threat, the FACT is that one day governments are going to start USING their "internet off switches." -It's the only way they'll be able to stop the information flow (and bitcoin flow) between people conspiring against them... They will not be able to shut down a mesh web, only take out key links that will regionalize it for a while.


#4 is why I'm excited about mesh. P2P software is the most important tool in the world for preserving freedoms, and governments are going to have to face it head-on soon. That means cutting off the web.

[benjamindees]

Price negotiation is 'expensive'

No it isn't.  Just maintain multiple links.  Periodically drop and replace the most expensive / least reliable one.

I don't really think meshnets are achievable as the anarchists would like them.

I agree, but for none of the reasons you stated.  They aren't achievable because too many "anarchists" will only settle for free internet and not "internet at cost."

Within the next few years, geosynchronous cubesats will open the door to internet access anywhere on Earth.  Either a Bitcoin-based meshnet will be ready to take advantage of them, or they will be snuffed out by commercial interests.  That will probably be the closest that humanity ever gets to truly "free" internet.

[BTCLuke]

They aren't achievable because too many "anarchists" will only settle for free internet and not "internet at cost."

You're clearly not seeing the other benefits to us anarchists. We'd pay 10x our current ISP fees for the freedom a true mesh would give us... It's the sheeple, people used to receiving government handouts, that will only want the free option.

Within the next few years, geosynchronous cubesats will open the door to internet access anywhere on Earth.

Geosynchronous? You do know that's going to take a bit more money to launch it that high, right?

That'll take some serious moulah. In todays' dollars, a million bucks worth of payload delivery to LEO would be several Billion dollars in price to deliver the same payload to a Geo orbit... Very few rockets on earth right now can even get up there!

[dserrano5]

1. Free internet. (As in zero-cost) There are tens of thousands of people using their mesh networks in Europe and elsewhere today as their only way to get online; and all they paid for was the hardward costs... No ongoing fees of any kind.

2. Free internet. (As in liberty) Don't like how you ISP is stopping you from downloading torrents or porn or something else they feel is wrong? Sorry, obomba told them to do it. So why should you have to go through the trouble of encrypting things like torrents when you could just do away with ISPs totally?

Unless I'm missing/misunderstanding something, the local mesh has to connect to the internet at some point, through someone's ISP, which would render these 2 points void.

[BTCLuke]

Unless I'm missing/misunderstanding something, the local mesh has to connect to the internet at some point, through someone's ISP, which would render these 2 points void.

What you're missing is that eventually, the mesh would BE the internet, and the existing internet would go away.

Meanwhile, if only one person in 100 or so on the mesh volunteered to connect their mesh router to an internet connection, they'd be able to share their connection to the existing web with all of us... And looking around in this particular community, I think the number of volunteers would be very plentiful indeed.

[greBit]

but for everyone else what will it offer?

4 things:

1. Free internet. (As in zero-cost) There are tens of thousands of people using their mesh networks in Europe and elsewhere today as their only way to get online; and all they paid for was the hardward costs... No ongoing fees of any kind.

2. Free internet. (As in liberty) Don't like how you ISP is stopping you from downloading torrents or porn or something else they feel is wrong? Sorry, obomba told them to do it. So why should you have to go through the trouble of encrypting things like torrents when you could just do away with ISPs totally?

3. Small interruption avoidance. If a truck or lightning bolt hits into the wrong telephone pole up the road from you, you'll be without internet for hours or days. Not so with a meshed web.

4. Large interruption avoidance.  As little as you personally think this is a threat, the FACT is that one day governments are going to start USING their "internet off switches." -It's the only way they'll be able to stop the information flow (and bitcoin flow) between people conspiring against them... They will not be able to shut down a mesh web, only take out key links that will regionalize it for a while.


#4 is why I'm excited about mesh. P2P software is the most important tool in the world for preserving freedoms, and governments are going to have to face it head-on soon. That means cutting off the web.

I do not believe that many governments would commit economic and political suicide and risk popular revolution by cutting off the internet. If it ever got to that stage I would probably be more worried about not getting blown up in the street than about my internet connection.

I agree about the benefits of using Meshnets for niche applications - especially in areas where ISPs are too expensive/unreliable/unavailable or where people are genuinely afraid of government cutting their connections. So I would love to see cheap and easy Meshnet kits that would serve this demand.

But for the majority, just swapping out the connectivity layer isn't going to magically bring them freedom as in liberty on the internet. I can achieve this today over my existing connection. I can connect to a P2P darknet, emulating a 'meshnet', and no ISP or government can monitor or censor what I publish and access online.

The problem is that the current tools are clunky and only used by a tiny tiny minority.

One of the first things we need is an efficient distributed datastore, with anonymity guarantees and censorship resistance. Thanks to Bitcoin, (freenet/I2P/Tor)-like protocols can now be augmented with market-based incentives to encourage adoption. (something like Bytemaster's tornet - https://github.com/bytemaster/tornet)

Even with a simple anonymous datastore primitive, many things can be achieved.

• Decentralized email. Secure. Wire-tap proof
• Alternative to Twitter
• Discovery services & marketplaces (ebay/silkroad/...) could work over this. Open source app with dumb cloud back-end
• Youtube replacement
• etc etc

It is my opinion that one can achieve freedom on the internets not by trying to replace the ISPs but by gradually taking the power away from the large centralized services and making the internet decentralized once again.

There are already positive signs as we see that more and more processing happens on the client-side. Javascript based apps that interface with some API in the cloud are becoming the norm. I would like to see this trend continue, a dumbing-down of the cloud with richer clients.

[ElectricMucus]

1. Free internet. (As in zero-cost) There are tens of thousands of people using their mesh networks in Europe and elsewhere today as their only way to get online; and all they paid for was the hardward costs... No ongoing fees of any kind.

2. Free internet. (As in liberty) Don't like how you ISP is stopping you from downloading torrents or porn or something else they feel is wrong? Sorry, obomba told them to do it. So why should you have to go through the trouble of encrypting things like torrents when you could just do away with ISPs totally?

Unless I'm missing/misunderstanding something, the local mesh has to connect to the internet at some point, through someone's ISP, which would render these 2 points void.

The larger the mesh becomes the less it has to rely on the existing architecture. Second there is not just one point where the mesh could connect to the rest of the Internet. Any node can function as a gateway if necessary.

[blisscan]

I've tried qmp.cat

See map.qmp.cat for a map of existing nodes .

[benjamindees]

Within the next few years, geosynchronous cubesats will open the door to internet access anywhere on Earth.

Geosynchronous? You do know that's going to take a bit more money to launch it that high, right?

That'll take some serious moulah. In todays' dollars, a million bucks worth of payload delivery to LEO would be several Billion dollars in price to deliver the same payload to a Geo orbit... Very few rockets on earth right now can even get up there!

It will be expensive, no doubt.  But cubesats are really small.  Cost will ultimately depend on whether they can piggyback aboard another launch:

Through companies like Eurokot and Kosmotras, the launch costs are currently about US$40,000 per single cube.

http://www.satmagazine.com/story.php?number=602922274