Automatic Wireless Network Organisation

babycakes writes "Interesting article over at the BBC about the future of wireless networks. Researchers at Intel are developing mesh networking technology that automatically organizes the best routes between wireless devices for better bandwidth control, now that the number of wireless devices is set to explode." Neat stuff, but they've yet to implement any sort of security controls. Until they do, it's going to be a real party for wireless eavesdroppers. timothy adds: La Camiseta writes "Wired has an article discussing the extent that some people will go to in order to stay connected. My favorite is their 4-way SMP system that fits into a briefcase."

Comments

Security implications?

[tjansen]

I wonder what the security implications are. If every device is able to route, a malicious device could claim to have great connections to other devices in the mesh and then drop packets. Unless there is some way of authentication in the mesh (so that only authenticated devices can participate), it would need some trust/rating system so devices can exchange information about the reliability of other devices...

Re:Security implications?

[Anonymous MadCoe]

I was thinking about leaving a trace of where you've been (just like your mobile phone does right now).

Re:Security implications?

[Anonymous Coward]

So you expect people to reply to you when you talk, but they're not allowed to know where to reply to? Care to explain how that works?

I guess I'm slow...

[mao che minh]

So they intend on doing away with the traditional hub/switch/router layout, in which someone administers the network and decides what amount of bandwidth goes where, and instead embed these functions into the wireless devices themselves? Seems a bit ambitious.

Whats the point of that Mini-ITX cluster?

[stratjakt]

I have to assume those are the 800mhz versions from the CPU fan, but still, these things have very little power from my experience with them. Cool little toys, maybe enough for a low-end workstation or terminal, but I wouldnt think of using one in a server situation.

I actually used one to run my squid-based proxy, and it was outperformed by the P200/MMX I wanted to replace.

Why cluster 4 of those? I'd think a 2 gig Celeron (of course P4 would be preferred) would have more power, faster DDR Ram, be as cheap (if not cheaper).

I cant fathom why one would cluster those little Eden boards, unless they were going for redundancy; a poor mans fault tolerant server.

Anyone?

Re:Whats the point of that Mini-ITX cluster?

[bheerssen]

Two words - power consumption.

Those little Eden boards consume very little power. Clustering them would provide more robustness and processing power for a mobile network.

Old Hat

[NimrodMCSE]

The company I work for has been doing this for over 2 years. Seems like Intel is way behind the curve on this one. You can see what we've been doing @ http://www.meshnetworks.com .

Re:4-way SMP?

[stratjakt]

If they're all working for the same task, like distributed compiling, or just running the same OS, then I think it can be called SMP. If it's just 4 redundant servers, I'd say not.

I dont think SMP means all the procs need be on the same sheet of fiberglass.

Then again, I may be wrong and will be shouted down by some nitpicking geek.

Wireless security IS a big issue...

[DarklordJonnyDigital]

...when the modern nerd can hack from low-flying aircraft passing over your house! Ars Technica have done it already - twice, no elss: War Flying, 1 and War Flying, 2

This is just too cool.

Re:4-way SMP?

[AssFace]

exactly - I was going to say there doesn't look to be any use of shared memory (not even sure how they would).

I'm curious if they are writing stuff with something like PVM for a beowulf type thing, or the more likely OpenMosix option.

Also curious as to what they are using it for. (didn't read the article but checked out that briefcase b/c it sounded and looked cool)

Re:Battery usage?

[vadim_t]

Why not? You get something in return for the resources you have to give up: a network.

It sounds like Freenet to me. In order for the network to work well people need to run nodes. Yes, they will transfer stuff for other people, something that doesn't benefit your directly. But it also keeps the network ticking, which is what you want.

I suppose it will work this way: If you want the advantages of the network then you have to contribute with your battery.

Security in "cooperative" networks

[per unit analyzer]

The problem with security in ad hoc mesh networks built in a cooperative environment* is that you have to worry about who's running the intermediate nodes and whether or not they're doing anything bad with your traffic. Even if you design the wireless link so that non-participants can't sniff your traffic out of the air, the mesh nodes themselves must be secure so that their owners can't snoop on each other. So in the mesh environment, you'll either need to require security to be end-to-end or design a radio that switches packets at a low level and doesn't let traffic passing through to be copied higher into the node OS where it can be snooped. I would opt for the end-to-end security since I can't be sure someone hasn't built a radio that violated a privacy standard.

--zawada

(*) By cooperative environment, I'm talking about mesh networks that are built by many independent entities, with each node carrying the traffic of multiple participants. e.g., your laptop participates in a cloud of laptops that ultimately carries all participants' Internet traffic back to an access point somewhere.

Mesh Routing algorithms

[TheSync]

Mesh Routing links.

MeshAP/MeshBox from LocustWorld is based on AODV routing module. There are also userspace versions of AODV. Plus there are at least two other competing mesh routing algorithms, and there are other hardware mesh solutions, but MeshAP has a lot going for it.