Slashdot Mirror

We run a uClinux variant (open-ap) on wireless access points too.

It's good to see it integrated into the kernel though - I can use the latest firewall and access control features on my little accesspoint - and if it won't fit? I've just got to pick something else I can do without.

I see they are working on a port to Cisco 2500 routers too - though it is a long way off, and doesn't look like it has had mucch recent work. There is also a port to the processor in my old Cisco 1003 - when that all gets merged in (probably years away) we will have something I can't even get from Cisco - a current operating system for old hardware (they stopped supporting the 1003 back at version 12, and I wanted a 12.1 IOS feature - VPDN (yes, there is a linux project, but that needs some more features before it goes into my 1003 :)).

The impossible we can do next week - miracles may take a little longer!

Your project sounds very similar in needs to another application I'd discussed with a friend some time ago. When the Civil Air Patrol gets called out on a search-and-rescue mission, they frequently fly over the suspected area with a video camera, sending frames down to the ground with slow-scan TV so that an expert on the ground can identify likely spots to search. The image quality sucks and the data rate is worse.

It'd be nice if these folks could carry a multi-megapixel digicam on the plane, snap pics of anything interesting, and have the images sent to the ground for viewing/zooming on a laptop. The only trouble is, the search area is usually several miles on a side, well over the range of normal 802.11a/b. The solution we came up with involves a cadet on the ground with a high-gain dish antenna being told "keep this pointed at that plane". Cheaper and more reliable than an automatic antenna tracker, for sure. :)

So what we need here is a way to interface with the digital camera. As soon as a photo is taken, we should suck it of the camera's memory and buffer it for transmission to the ground station as soon as possible. Point to ponder: Assume that wireless connectivity is intermittent. Do you transmit the most recent pictures first, or the oldest pictures first, to make sure the base station has the most useful data possible? (LIFO or FIFO?)

Some time ago, I proposed a bluetooth CF module which would appear as a large FAT filesystem. It would have a limited amount of "cache" memory, where images would immediately be written, and then it would then link to a large hard drive sitting in the user's pocket and free the cache for more images. An 802.11b version would suck more battery than bluetooth, but allow longer range operation. The trick isto emulate a filesystem, so firmware hacks aren't needed. (If you could mess with the camera's firmware, you could use an existing CF wireless card and let the camera handle the protocol, right?) Nikon seems to have released a similar product but it's vaporware and only works with one particular camera.

In lieu of sitting straight in the camera's media slot, there's always software that controls the camera via serial or USB. The problem is, most of it seems designed for interactive use. I don't know whether the protocols support lurking in the background to just suck files off the flash card, without interfering with the camera's normal operation. Several of the packages are based on a common code base with a protocol that's fairly well documented, so rolling your own isn't out of the question. Let's assume for the sake of discussion that your camera supports this and suitable software can be found or written.

If you can get by on RS232's peak speed of 115200bps, and if the software can be worked out, there are several hardware options. Several other posters have suggested PDAs with wireless cards. That's a great idea, especially if you can strip off the screen and case to save weight. Some suggested the Soekris net4511 or similar. It's got a low-power 486 chip, serial ports, ethernet ports, and a PCMCIA slot for your wireless card. Or, you could hack up an existing 802.11b accesspoint to run linux and use its console port to connect to the camera. (Note: The Eumitcom-based APs are getting hard to find now. Not a platform with future potential.) There's a similar project for the Apple Airport base station, but it's limited to etherbooting, probably not suitable for this application.

You could use a pair of Ricochet modems dialed to each other (auto-answer on the chopper, and dial from the ground), in which case they simp

Is the protocol that enables mesh networking an open standard? The OpenAP project and the Mesh AP folks use 802.1d, or MAC bridges, to eliminate redundant hops.

The reason this is useful is that it allows a group of (perhaps constantly changing) network nodes to form a big cohesive network. It sounds like it creates isolated layer 2 cells and builds a single layer 3 network out of the cells, which I believe is what the Linksys WET11 does.

Compare this to the coherent layer 2 network which can be created with Spanning Tree Protocol by using OpenAP like these guys do.

Most WiFi access points do not support a mesh topology, but only support hub and spoke. With hub and spoke, you can only connect to the network if you have line of sight to a hub, but with a mesh network, you can connect via any other network node. Mmmmmmm. . .

Perhaps someone who actually got through the slashdotting can comment on the other features (compression, encryption, proxying, etc.)

This thing runs linux. So it must be better.

Linux based Access Point

First, if you're going to insist on a REA-type federal approach, at least insist that a hard time limit be imposed on the agency. The REA hung around until 1994, and even then was assimilated into a similar government agency. Bureaucracies don't ever put themselves out of work.

Second, does it really take that much money to get started? And that cost is continuously dropping, especially with the help of open source projects. Isn't it likely that this problem can be solved within the next several years without government intervention? I'll bet on private initiative and creativity (hmm, solar powered WiFi relays anyone?) to get the job done.

Besides, if taxes weren't so damn high more individuals would be able to finance such initiatives. It would be far easier to bootstrap ventures if, say, we didn't have 1-1/2 months per year of labor confiscated by the Socialist Security system.