Slashdot Mirror
Meshcube: A New Mesh-Routing Wireless Device
jazzgroove writes "The Meshcube is a new wireless mesh routing device based on open source technologies. It's quite feature rich with support for VPN and IPSEC which come from the meshcube distribution. Apparently you can buy the device as a kit and build it yourself or buy it pre-built. For more information have a look at the wiki."
"Especially for communities we offer the meshcube as a kit which can easily be assembled. Prices start at 199,90 EUR. The kit comes without antennas to allow highest flexibility to the lowest possible price."
= $241.98
"For those who want to start meshing right away this true linux-device can also be ordered readily assembled - prices starting at 239,90 EUR."
= $290.62
Calculated as of this morning.
Hmmm.
right now the contents of the FrontPage are:
eden.h4xx.com - whacky free for all image board
It's an access point that can be joined with other access points to form a "Mesh". So for example you can put a bunch of them in a town and share one or two internet connections to people who dont have such a connection. Also useful for networking over distances that one single AP wouldn't cover (by chaining them together through the airwaves).
Jon Anderson pioneered this with his boxen, see http://www.locustworld.com/
64 MB Ram, PoE (802.3af), smaller dimensions, 400 MHz Mips CPU ..
... there are a bunch of pricewatch entries for the WRT54G at under $70.
More importantly, an optional second Wifi interface: One for a backbone, one for clients.
PS: The 199,90 price given is for a kit. Assembled it is 239,90.
Seastead this.
It's an embedded Linux box with miniPCI ports, a processor and a distro on it. What you can do with a (slightly vintage) PC and a bunch of PCI WLAN cards, you can do with a Meshcube and a bunch of miniPCI WLAN cards, provided that you can shoe-horn your software into the provided 32-64MB of solid-state storage, or are content with an external USB-attached hard drive. I guess the correct answer then is "router", "AP" or "bridge" depending on what actual software you put on it.
It seems to have much, much much more room than a WRT54G/GS, but it sure is quite pricier (the Linksys units are around EUR 90 and 110 incl. VAT respectively, though the GS is only now supposed to come out of pre-order).
From the energy point of view, 4W for all they say the MeshCube does certainly looks very nice.
* 400MHz MIPS processor AMD Au1500 aka Alchemy
* 64 MB RAM
* 32 MB Flash
* 100Mbps Ethernet
* Power Over Ethernet Standard IEEE 802.3af
* USB host
* USB device
* up to 8 MiniPCI devices, the base unit comes with one dual adapter
* WLAN cards with RP-SMA connectors
* Small Size 7x5x7 cm
* Low power consumption 4W
* No moving parts
* one (hidden) DebugConnector with serial port and EJTAG
Its pretty sad when a commercial OS ships a debugger with their system but no compiler.
MeshCube OpenSource Distribution
This will become the main site for the MeshCube OpenSource distribution. (Until we have set up our webshop at http://www.meshcube.com, we also abuse the domain meshcube.org for hosting our order form.)
The MeshCube is a new hardware platform dedicated to WirelessLAN mesh routing, developed by [http://www.4g-systems.biz 4G Systems, Hamburg]. With a 400MHz ["MIPS"] processor, 64MB RAM and 32MB flash, and up to 8 MiniPCI cards, it is powerful enough to provide excellent security and encryption, and flexible enough for custom applications and modifications. See http://meshcube.org/english/specs.html or HardwareSpecs for more details about the hardware.
The MeshCubeDistribution is the Linux distribution running on the MeshCube. Its main features are MeshRouting, autoconfiguration of networking, an emphasis on security (IpSec, VPN), and a compact design (to fit on the 32MB flash). It is completely licensed under the GPL and will be developed here in our CVS in true Open Source manner. We happily accept patches and additions, but please be patient -- it takes some time to evaluate patches and import sources into CVS.
We have set up several mailing lists for communication related to the project. Please subscribe here. We've also set up a Bugzilla (bug tracking system); please use it to report any bugs you may encounter.
There is a feed of packages you can use to install new software on your cube and to update your installation. See HowTo/InstallPackages for instructions.
This wiki is free to edit for anyone and currently divided into four sections: HowTo, MeshRouting, HardWare and SoftWare.
Frequently Asked Questions
CVS
Mailinglists
Bugzilla
DownLoad
I have not found any hint on this wiki to an IRC-Channel for people which uses the meshcube, so i opend one. madd.
* Server:......irc.freenode.net
* Channel:...#meshcube
Hmmm.
Although I've been admiring Soekris boxes lately, I don't think they are a good fit for this... You can only plug in one Cardbus/PCMCIA card, and a MiniPCI device if you want, and that's $200... Meanwhile, the MeshCube says it has room for 8 MiniPCI cards...
The Soekris boxes only have low-end processors, in the 233MHz range... A 400MHz MIPS processor should be faster (not sure of that), although x86 compatiblity is a good plus for Soekris.
The MeshCube claims to have (slightly) lower power consumption than the Soekris as well.
Slashdot gets worse every day... Pipedot: News for nerds, without the corporate slant
The meshCube is excellent for mesh netwroking using MANET routing protocols. It has sucessfully ran the optimized linkState routing protocol from http://www.olsr.org for self configuring multi hop operation.
This has been done. OLSR(http://www.olsr.org) packages are avalible for OpenWrt(http://openwrt.ksilebo.net)
along with the broadcom chipset
Unless you are a fan of broadcom or something, I don't see how this is necessarily a selling point.
This meshcube does have a pile of miniPCI slots which I think would be useful for many point-to-point relays to set up a mesh (one for service, four for connecting to adjacent lattice points) in contrast, a similar setup using that model Linksys would require five of them.
Because Mesh can reduce the cost of proliferating wireless access over any area even and especially in a building and provide for high levels of administration. Consider:
-Mesh nodes have the power to be aware of other nodes and auto power so they don't cause excessive interference with each other.
-You only have to deploy one piece of equipment to spread the signal instead of running wires to the new location and/or having a wireless bridge connected to an AP.
-It is scalable and upgradable. You only need swap wireless devices or add in a new one to take advantage of new wireless technologies and/or have multiple frequencies in one device.
-They are completely flexible so that you can setup any network topology you want as opposed to off the shelf WAP's that are extremely limited in ability.
-The flexibility allows for amazing abilities to monitor and control all functions and activities on the APs.
-You control whether there are backdoor accounts.
That should get you started.
Stay FAR FAR away from Miktrotik. I could sit here and tell you horror stories about it that I have first hand but here are the highlights:
-Unpatched security holes known for YEARS. Run NMAP against a Mikrotik box and watch what happens (at least since the last time I used it).
-Unwillingness to provide source code to GPLed packages included (again since the last time I used it).
-Many failed upgrades rendering boxes usless despite following instructions to the letter.
-Unwillingness to refund money when new licenses had to be purchased for failed upgrades in the middle of the night.
-Poor documentation (it has gotten better)
It isn't all bad and does do some neat things but it is far surpassed in its usefulness by the introduction of Mesh technology such as this in my opinion.
I saw one of these last November at an event in London attended by a lot of the consume.net people. The photo does not do justice to how small it is, about 3" by 3" by 1". It can be powered up the cat 5 so it would be relativly easy to mount one on a roof.
The software is based on debian. Christian Car told me that "egiht hops and there is no bandwidth left" which means my dreams of city wide mesh networks are going to need somthing resembling network planning. The biggest obstacle I can see at the moment is that there are literally tens of thousands of access points in London. Access points transmit several bursts a second even when they are not transfering any data. It is possible to do five mile links between convinient points but even with high gain antennas the peers will still hear lots of access points announcing away plus all the other stuff in the 2.4GHz band.
The company is half a dozen geeks. They have great technology but don't seem to grasp the importance of documenting it and having a good website with lots of info on what it does and how to use it. I mailed them eight months ago say "everyone thought it looked really cool in the transparent case, at least put a picture on your website!". The biggest barrier to mesh network seems to be a people problem rarther than a technologial one. Getting a mesh net running requires alpha geek level network and computer knowledge. We have enough alpha geeks. We need more people who can recruit people and get access to sites for nodes and organise the geeks.
Looking at the Consume node database, it looks like coverage is hugely less than 5 miles from each node.
Most of the nodes in the consume database are not linked to each other. There are only a handfull of mesh network in operation.
The consume net people have started using a VPN over the internet to link nodes which of course is somewhat limited in bandwidth.
Or more specifically, how far is eight hops?
That up to whoever builds the network. It depends if you want blanket coverage or not. Mesh networking could be used to cover a group of nearby building where it is hard to get cables between the buildings.
You need the antennas high above obstructions to do long distance links due to the fresnel effect. A five mile link needs antennas 40 feet above the rooftops of the buildings in the way. You might be able to make a 40 mile long chain of mesh routers but you don't get five mile radius coverage arond each router. Another radio with an omnidirectional antenna will provide coverage for peers within a couple of hundred meters of the router. A node in the link path four miles from each router needs a high gain antenna on a rooftop.
For an alternative, check out the linksys wrt54g and the wrt54gs.
Linksys has released the full source code and build environment to these routers enabling you to extend the linux install that these are based on.
These devices are not as powerful as the meshbox, but they are much cheaper. They also have more ethernet ports to add a switch/dmz/whatever, although there's only a single wireless device.
Multiple projects have sprung up to extend the functionality of these routers, along with a bunch of informational sites about other hacking projects
Sveasoft continues to push ahead with adding features and have discussed adding mesh functionality as long ago as January of 2004. If they could do this then it would extend the market for the Linksys WRT54G(S). This is all another example of how the GPL can benefit everyone: consumers get cheaper, fully-featured wireless routers, Linksys/Cisco get a larger market. Good thing that Rob Flickenger forced Linksys to live up to the GPL.
Sveasoft mesh discussion