Meshnets on Maidsafe


#1

Maidsafe fundamentally decentralizes the net. It allow everyone to store a portion of the net on their computer, it decentralizes the way we organize the net, it’s secure and encrypted, and so the concept of a mesh network seems rather natural for me. Every computer is capable of acting as a server, everything from a cell phone to a desktop computer. If a computer could log into maidsafe and participate in the maidsafe data economy why couldn’t they act as servers in a mesh network? And AS a mesh network the SAFE network would become less and less relient on the centralized internet and more and more secure. So I think this an important feature to build into maidsafe, either in the actual code or in an app form somehow.


Mesh network integration?
Why sigmoid? + Safecoin crowdsale generation
SAFE and Mesh Networking
#2

Interesting point @Blindsite2k

You can run a meshnet overlaid on top of MaidSafe, which on its turn can run on top of another meshnet. MNoMSoMN :slight_smile:

It would be really interesting to run (at least some portion) of the Maidsafe network on top of a user-controlled mesh network infrastructure, rather than just as an overlay on top of the existing commercially owned Internet infrastructure.

For example, MaidSafe running of top of a cjdns meshnet would be pretty cool. If you want to learn more about cjdns, check this interview with Caleb James Delisle (I particularly like this excerpt at 10:09).

Edit: David discusses this subject in meeting #1.


#3

I don’t think my opinion is going to be a popular one.

The idea of decentralized mesh networks taking over the infrastructure of the Internet sounds nice. Unfortunately, with the technology we have now, it is not feasible. The problems are fundamental issues with the physics of moving bits around.
At the end of the day, basic telecommunication is about providing high bandwidth and low latency. Customers have an insatiable need for both and they will always strive to get more bandwidth and less latency. Centralized, wired networking infrastructure will always win. Whatever the level of technology we have, centralized systems will always provide more bandwidth and less latency for the same price.

Mesh networks are amazing technology, and they have their place in specialized niche applications and scenarios. They just do not scale to the level of backhaul Internet traffic.

Bandwidth:
Decentralized mesh networks imply wireless. The amount of communication traffic that can be transferred across a given area is limited by the radio spectrum available. Any transmitter broadcasting in the same area will have to share whatever maximum limit is available for the spectrum being used. High speed wired connections user fiber optic cables. These cables have the same limit, there is a theoretical maximum amount of data that can be passed down a cable. The advantage is, you can always string another cable.
With wireless, once you saturate the spectrum in an area, you are done. With wired, you can have as much traffic as you are willing to string cables to support.
Any advancement that improves our ability to more efficiently use wireless spectrum can generally be applied to fiber optic communication as well. Wired communication has always provided more bandwidth than wireless, and there is no research that indicates this will change anytime soon.

Latency:
In the digital world, latency is all about the number of hops your packets take, and how efficient your routers are. We can assume the routers in both schemes are equally efficient, therefore latency is all about the number of hops your packets take. A centralized system is architected to reduce latency. This is why we have giant backbone connections between major cities. These are the fast lanes that get your data as close to your destination as possible. With mesh networks, you have a large network of nodes, but no major trunks. There is no fast lane. Due to the fundamental design of a mesh network, you will always require more hops to get from one place to another, compared to centralized systems. It is the difference between driving on the highway system, and trying to get everywhere in the country using back roads.

There is also the problem with coverage. Mesh networks need a path from any location in the network to another. This might be easy in a city, but it gets very spotty between cities, and it becomes impossible across oceans. If we start talking about satellites, we have an even worse problem. We now have to deal with the time it takes light to get from the earth to the satellite and back. The round trip for light to a geostationary satellite is 250ms. Common satellite internet connections have latency on the order of 600ms. This is unacceptable, unless you don’t have any other choice. I have lived in rural areas. Trust me, it really sucks. Also, satellite communication is wireless, so you have the bandwidth problem described above as well, and there will be A LOT of people sharing that single pipe into your satellite.
LEO satellites would be much better off, with latency from transmission at around 40ms. This wouldn’t be horrible, but LEO has a whole slew of problems that require technology and coordination capabilities that are not currently possible. Either way, even LEO will always be slower than terrestrial connections that don’t need to throw their photons to space and back.
All of that fiber strung across the earth and the oceans is there for a reason.

Now, here is where you’all really start to dislike my post:
I predict that, if Maidsafe is successful, the network will trend toward becoming increasingly centralized. Large farmers will take over and be the dominant force in the network, and this is a good thing. I want a copy of the data I am looking for to be as close to me as possible, that means being in a large datacenter, in my city, with a fat pipe. I also don’t want my cell phone serving data, it has limited battery life and limited bandwidth (that wireless thing again). I would rather run a large server at home, that I can ensure has a fat pipe that will nearly always be connected. No matter how far technology advances, my cell phone will always have less power, storage, and bandwidth compared to my home computer.

You can create systems that limit the forming of large farming systems, but this will only limit the network. Centralization will lead to higher bandwidth and lower latency. For a fundamental infrastructure, like the SAFE network, bandwidth and latency are the end all and be all of success. If we do not allow this to happen, someone else will copy the network with these restrictions removed.

In my opinion, trying to “decentralize the Internet” is not the promise of the SAFE network. The true power of Maidsafe is the further decoupling of infrastructure from functionality. Currently, if you want to build a modern application, you need to provide servers, bandwidth, storage space, etc. It also means that application builders have control of those things as well. By moving those tools down into the infrastructure, you not only free the builders from needing to build it all themselves, you also free the users from the control that such a design makes possible.
Maidsafe forces neutrality on the infrastructure providers, they don’t control what is run on their systems. There are rigid expectations on what resources they provide and how they do it. It is incredibly difficult for them to have any effect on the actual functionality built upon what they provide. And if they attempt any shenanigans, they risk bringing down the entire system.
The SAFE network won’t free us from centralized infrastructure, it provides a new social contract that dictates our expectations from that central infrastructure. We currently live in the monarchy of the interwebs, maidsafe is a democratic republic.


#4

I actually agree with you @oillio, to some extent even with the unpopular bit, in respect to the trend towards centralisation of farmers (similarly to what happens with Bitcoin).

I really enjoy my 250Mb fibre connection to the Internet and I don’t think I’d replace it any time soon for a multi-hop flaky wireless connection. This is why I was careful to add the “(at least some portion)” on my post! :slight_smile:

It is definitely a niche use-case, but nevertheless a very interesting one to try. Some people are really keen on removing the ISP from the equation. The interesting question is how the MaidSafe network can adapt to these slower nodes in such a way as to not penalise those who do not care for meshnets and want low latency and high bandwidth. Maybe opportunistic caching is the key here (as David mentions on the above video).


#5

@oillio @ktorn

These restrictions are going away. There is a new radio tech who I forget the name of which effectively eliminates base stations and makes individual devices able to deliver a vastly more capable radio network than the current centralised approach.

It is proprietary tech afaik but who knows, maybe it will be usable evenso.

Like MaidSafe and bitcoin, the tech for this radio network is actually much much bigger than what we first think, since it can not only carry data, but also act as a power grid.

I know that it sounds daft to - use mobiles as a power grid, but this is his nature works - all life on this planet is one big networked solar energy capture and distribution network. With decentralisation tech, man is about to make a big leap, long exploited by evolution.


#6

I would also point out that I said that meshnets should be supported not replace conventional internet. One should be able to choose to use either/or or both. Consider the following scenarios:

  • You’re in a small town. Lots of locals with computers but your connection to the outside internet sucks. Would a mesh network not sound more practical in this case, even in a supplimentory situation? Put maidsafe on top of that and it would be REALLY useful.

  • You’re out of range of your local ISP but you’re close to someone who is in range, or close to someone who is close to someone, etch etch,. Why not create a mesh network until someone connects?

  • You don’t want to be dictated to by an ISP to be part of a network. So you start forming a mesh network. (political activism angle.)

  • You want an extra layer of security (security angle)

  • You don’t want your maidsafe node to be taken down due to your ISP so you set it up with meshnet capabilities. Or you do it just so you can conect to the safe network even though you can’t get on the internet.

Have I adaquately made my point yet?

Yes sometimes mesh networks might have latency and bandwidth issues but that doesn’t always matter. Slow internet is better than no internet. Also with mesh network support people would be encouraged to build their own networks. And isnt maidsafe built to automatically find the shortest route between point a and b? So that latency issue is already accounted for somewhat.


#7

Thank you for posting this. It may be unpopular, but necessary to be realistic in what works. At the end of the day, people will decide between speed and privacy, control and freedom. As you have stated below…

This is my favorite part!


#8

@happybeing I’m sorry man, but this sounds like snake oil. These are limits set by our current understanding of physics. A lot of very smart people have tried to do the same thing, and failed. If what you say is true, there are Nobel prizes in store for the creators. I hope it is true, but I am not going to put any expectations in success. Until I see the peer reviewed paper or the product on shelves in Best Buy. Even if they get it working in the lab, we are still talking 10 years to a product.


#9

Somewhat off-topic
I’m worried enough about possible health problems as humans live in an ever deepening and more turbulent electromagnetic soup.
Putting power into that as well needs careful consideration. Maybe I’m getting my open-toed sandals in a tangle over this but I’d need a bit of convincing that there would no added health risks involved with PoW (Power over Wireless).

Apart from the (alleged) fact that we will all die from brain tumours, this sounds pretty cool otherwse. Now do I want to live longer or get better hassle-free bandwidth? Hmm it’s a toughie…


#10

@oillio I haven’t delved deeply into this (or I’d at least remember the name! ) but I’m not naive either…

“Any sufficiently advanced technology is indistinguishable from magic [or snake oil]” ~ Arthur C. Clarke

The principles are within current physics, no Nobel prizes needed, and the tech is actually working and if my memory is not betraying me, being trialled by the mobile industry. It may still be years to deployment, but I think less than ten.

Power distribution would come later, but maybe still within that timescale, depending on the availability of suitable services which would require a more coordinated effort, from more partners, to roll out.

@Southside it’s a concern I agree, although I can see ways in which the tech might either make the problem worse, or the same, our in fact better. Not that there is evidence for a problem yet - though I too remain skeptical about “no evidence for harm” claims for anything commercial, not just radiation.

Both of you: the tech works by having devices use a novel technique to focus transmitted radio energy into a narrow beam, rather than send it out in all directions. This combined with the ability to select and direct the beam only towards the most suitable receiver vastly reduces the power needed, and so decentralized (yes, that word again) communication becomes possible by routing through the devices themselves, avoiding the need to use centralised base stations and freeing up capacity many fold. It’s a similar principle to MaidSafe freeing up bandwidth and disk resources through decentralised networking and deduplication of stored data.


#11

You mention the physics upper limit aka Shannon’s Law quite a bit, so I thought I’d just briefly mention this new technology being debuted by the likes of none other than Steve Perlman of Apple, WebTV, Onlive, Mova fame. He calls it PCell or DIDO (Distributed Input Distributed Output) Wireless Technology and he’s built a company called Artemis around it.

He explains what it is better than I can so definitely start here:
http://www.artemis.com/

My thinking is though, that a LEO Constellation plus a terrestrial grid of these radio antennas coupled with a MeshNet protocol would provide size-able advantages in bandwidth to existing wireless infrastructure. Another possibility is having this as the backhaul for it’s own co-Internet, and not necessarily relying on the commercial Internet unless need be, in which case a connected node would then route it’s signal. The bottom line is this… Not everyone has fiber. Most people have cable modems with speeds of 10-30 mb/s Down. Whereas one node connected to fiber directing traffic over this new wireless infrastructure could allocate enough bandwidth to provide FAR higher throughput than that to the end user… I mean the guy is streaming 4k for Christ’s sake.

This new PCell technology is essentially the most disruptive thing to ever occur in wireless, it’s not something to be brushed off. Apparently they’re in talks now with major telco’s to deliver ASAP.


#12

Yes, thank you @RealHeadWay, PCell tech from artemis is the one.


#13

Check out this demo of the Artemis pCell Wireless. I don’t like hype, but it got me excited.


#14

huh. Trying to find more on this… this seems like a huge deal if legit…


#15

wow…just what we need


#16

@Blindsite2k Opengarden.com can help a lot to create an meshnetwork.

@ktorn CJDNS is not opensourced and Caleb James DeLisle’s idea to have a centralized DNS is totally WRONG. Namecoin could be used, but i got a few problems, if Namecoins become expensive it will be expensive to register a domain. Don’t get me wrong I love CJDNS, but it needs to be opensourced with decentralized DNS

@dyamanaka thanks for sharing


Taking decentralized one step further... wireless?
#17

This is really quite exciting. The use cases for this are out of this world. This, I dont think just can, will usher in the era of meshnets. I just wish the physics werent so far above my head. I want to know how it works.


#18

I have been following Artemis and it looks interesting. However, this is a centralized client/server technology, not peer to peer. It cannot be applied to mesh networks.
The technology requires a central server where all data going to/from the clients must pass. For download data, the server takes the data streams being sent to all clients in the area and transforms them into a series of signals, one for each base station in the area. Uploading data is reversed, the server takes signals from the base stations and transforms them into data streams.
Communication can only happen between the client device and server, through the base station. Client devices cannot communicate with each other without going through the server.
If this technology is successful, it will actually make mesh type networks less attractive, not more. It will give central wireless networks a giant boost in price to performance ratios, further outstripping similarly specd mesh network configurations.

Having said that, I am still sceptical. Artemis makes some heavy promises. They have demoed a few devices working in laboratory settings. Scaling to tens of thousands of devices in the real world may pose serious challenges. I hope they succeed, the technology would be truly game changing. I am not holding my breath, however. There are a long line of technologies that have performed flawlessly in the lab, but fail to be practical in the real world. I’ll get excited when their first large scale production deployment is successful.


#19

I’m well aware of the central data center that computes the algorithm for proper wave propagation amongst interfering peers on the network, thus allowing this sort of wireless “virtualization”, but it might be possible to use maidsafe in place of that central server, making it decentralized. The idea with this, is instead of back hauling to the main internet, for this co-internet, you use the nodes on maidsafe to compute this inverse interfering wave propagation simulation, and broadcast that to the clients from maidsafe. Yes, maidsafe itself does obviously have to been part of the main internet or some other pre-established infrastructure, no-one is saying this infrastructure could serve as a full-on maidsafe backbone as you run into a chicken and egg problem, it’s a logical paradox, a node simulating the algorithm cannot be one that gets the wireless signal to itself, that’s impossible, there does need to be a further backhaul in that respect. So… Potentially you could have a parent maidsafe and child maidsafes who have some type of QoS agreement, possibly with different arrangements but still relying on the main protocol for data relay. The bottom line is tho, for data storage and especially computation, these child MeshNet maidsafes with high enough QoS, are just fine for most developers and users of the system and users of the co-Internet service bypass the main internet by relaying packets through maybe just one node or a few nodes (those deemed safe) that are part of a maidsafe network.

In this approach tho, it would be ideal if nodes performing this wave propagation simulation are just dedicated entirely to giving you signal, and nothing else. In this respect you could maybe even using some combination of FPGA, ASIC, and GPU Hardware to create individual machines stationed nearby or completely encompassed (like a total package) by every single one of these radio antennas.

Bottom line, the data-centers only purpose is to calculate what amounts to an interference pattern, then send that to clients, leaving you with a clear signal. There’s nothing stopping any willing participant with some other pre-established connection from running this software, besides of course, it being Artemis’s intellectual property :wink:


#20

Opengarden only works with bluetooth and works better with wifi, it seems to be geared primarily towards mobile devices.