Homemade EVDO/WiFi Mobile Access Point 172
Tamundson writes "Over the last few weeks I've built myself a mobile access point for my car. It's based on a Soekris net2421 embedded Linux box and uses Verizon's 1xRTT/EVDO network as its uplink, resharing it over 802.11b. Wherever my car goes, my Internet link goes! :)
I finally put some webpages together on how I built it. The components are pretty cheap and anybody with basic Linux skills can build their own just as easily. I've also got it interfacing with Google Maps to do live vehicle tracking via gpsd. It also uploads pictures from an on-board webcam every five minutes or so."
Two, now four wheels (Score:5, Interesting)
JIC, coral links for website [nyud.net] and GoogleMap image [nyud.net]
...and back to 2-wheels again (Score:1)
Re:...and back to 2-wheels again (Score:3, Informative)
You could use the dual-axis accelerometer [digikey.com] I used for the robot project to sense tilt changes. Have it integrate changes to guess at current tilt, and rec
Re:...and back to 2-wheels again (Score:1)
Thanks for the parts pointer! *digs through Digikey*
stop moving! (Score:5, Funny)
Re:stop moving! (Score:5, Funny)
Now we can hold competitions: one AP driving around, several contestants trying to break into a machine in that car while they have to be in a certain proximity to have a connection to it. Sweet!
Re:stop moving! (Score:2)
Re:stop moving! (Score:2)
Lets see for the crew, we'll need a driver, a navigator, and backseat hacker.
Then we also have a weight problem. Normally people who drive cars weigh less than the average hacker.
Re:stop moving! (Score:3, Interesting)
Expensive (Score:4, Funny)
Re:Expensive (Score:3, Funny)
Re:Expensive (Score:1)
Why is it... (Score:5, Funny)
Re:Why is it... (Score:5, Interesting)
Building self-driving cars is a Very Hard Problem. It's being worked on, and great progress has been made, but it's not going to be ready for prime time yet.
The problem is that it has to work safely even under strange or pathological circumstances. Guaranteeing this is much, much harder than getting a car to drive on an empty road and stop at well-marked intersections.
On the plus side, as soon as a car autopilot drives better than the average driver, the insurance rate perk for getting one will make the switchover very rapid.
As for distraction, you'll note from the article that the access point was never used by the person driving the vehicle (and that it's in fact illegal to do so in California). It's a passenger perk (and great for when you get _out_ of the car, with the range it has).
Re:Why is it... (Score:1)
The other major problem in building self-driving cars is that unless everyone is using them, a self-driving car has to be able to share the road with human drivers, not just other self-driving cars (w
Re:Why is it... (Score:2)
Meanwhile, in the real world, you're driving on poorly-maintained roads with bad signage, sub-optimal weather conditions, and lots and lots of other drivers doing unpredictable and quite often illegal things.
Driving on a well-marked road under good conditions is a toy problem by comparison. IAACompEng, and a regular driver.
Re:Why is it... (Score:2)
Even having a car that drives itself on four-lane divided highways is beyond the reach of reasonable current technology. Don't even think about doing it in city traffic!
Sad, I know, but it's not as though no one's tried.
p
Re:Why is it... (Score:2)
Not to mention in open desert for that automated robot car competition they just had. None of them made it to the target distance.
Re:Why is it... (Score:1)
Re:Why is it... (Score:5, Interesting)
Simple things were serious issues like - if you're going slowly and your wheel comes up against a rock and at the current amount of throttle, it can't get over, what do you do - how do you know to just give it a little more gas and drive right over vs. you're up against something a little bigger that you should back up and drive around.
Re:Why is it... (Score:2)
Re:Why is it... (Score:2)
Re:Why is it... (Score:2)
No, we want to use them, we just don't want to pay the cost to make them work.
If it was one block from my door to the train, one or two switches total to work, and didn't take forever (ie. 3 hours to go as far as a car in 30 minutes) to make the trip I would take the train. However there is no way people on my income (and I make above average) could afford the ticket prices to have that level of convince. So only large cities where the population density allows sharing the costs between more people than
Re:Why is it... (Score:2)
Finally!!! (Score:3, Funny)
Beats the method of calling my friends every half an hour when i'm on a trip and trying to describe the surrounding cities/streets/lamp posts in hope for some guidance to a hot cup of coffee.
--Beware of on the road browsing though
Re:Finally!!! (Score:2)
You should be going here [starbuckseverywhere.net]. The guy's visited 4535 so far, or about 90% of the corporate Starbucks in the country. And others, abroad. If you haven't got a wireless setup for your travel, yet, you can visit the states beforehand and print/save pictures
Re:Finally!!! (Score:2)
Must be a "Living in the City" thing.
Slashdotted ... or? (Score:1, Funny)
Re:Slashdotted ... or? (Score:5, Funny)
Re: (Score:2)
Re:Slashdotted ... or? (Score:1)
*starts sloooooowly copying images over*
Pictures fixed (Score:1)
Seems like a good prototype. (Score:5, Insightful)
Re:Seems like a good prototype. (Score:3, Insightful)
I'm not sure how he'd get the cost down much further than it is already. He's using off the shelf parts at commodity prices. Building in quantity-100 might shave off 30% or so, but even fabbing your own integrated board with all widgets on it in quantity-1000 would only get to about half the cur
Re:Seems like a good prototype. (Score:2)
Don't forget that much of that is optional features. A trucking company doesn't need the WiFi card, or the webcam. They just want a display (controlled remotely) to tell the driver where his next stop is, perhaps with a few acknowledgment buttons.
Re:Seems like a good prototype. (Score:1, Offtopic)
-JF
I call shenannigans on that one. Ya figure, people nowadays are living 2 or 3 billion seconds (60 to 100 years). If each of the ~275 million Americans alive today only paid for 1 of the seconds of your life, you'd have to die by age 10 for the math to work out.
Then again, maybe the US gov't really *is* spending money 10 times as fast as it's coming in...time to cash out my bonds!
Re:Seems like a good prototype. (Score:2, Offtopic)
Assume that the average American starts working at 16 and retires at 55. This leaves 39 years of working. We can round up to 40 years of working years.
with an average of 40 hours wourked out of every 168 hours in a week, slightly less than 1 in 4 hours is spent working while employed.
Therefore the average person works for less than ten years of their life, which comes close to the number for how long you said
Re:Seems like a good prototype. (Score:2)
Re:Seems like a good prototype. (Score:2)
That 10% won't really change the fact that your calling of shenannigans was unjustified. I just showed it isn't that "you'd have to die by age 10 for the math to work out" so much as a per
Re:Seems like a good prototype. (Score:2)
I think the math could come close to working.
The only problem (Score:2)
And that's just not happening.
But google maps partially does this now, in selected cities.
Re:The only problem (Score:2)
And that's just not happening.
If one car in a hundred had a transponder, you'd get a good idea of how traffic was flowing on the major routes in and around a city. One in a thousand might even be enough for this.
You might actually be able to do this just by looking for cell phone emissions from tracking towers. I'm sure at least that many people use
Re:The only problem (Score:2)
It'll be for road charging - your road tax depends on how far you drive and where - but of course they'll use it for speeding tickets pretty soon afterwards, and the security services will have a back-door immediately, and then they'll let the police use it.
Tinfoil hat time for biglig, you think? Did you know that Mayor Ken Livingstone has publicly stated that if he ever canceled the congestion
Where do you get Verizon Broadban d (Score:1)
Re:Where do you get Verizon Broadban d (Score:2)
Verizon offers this in Austin, I have it . . . (Score:2)
The 5220 card runs 200 bucks with a 100 buck discount and the service is 80 a month.
I haven't found a dead spot in Austin yet and I know it is good as far down as San Marcos.
Re:Where do you get Verizon Broadban d (Score:2)
The problem isn't detecting traffic jams. They usually occur in the same places. The problem is that in most places, there is no way to circumvent the traffic jam. In the NYC metro area, most of the area highways are saturated during the peak-period, even without any anomalous events. If one road is diverted, where are the cars going to go? Local roads cannot handle the amount of traffic a highway can handle. Other highways are already satura
Slashdot effect in acction (Score:1, Funny)
Re:Slashdot effect in acction (Score:1)
Re:Slashdot effect in acction (Score:3, Funny)
Yeah, that will help him. All Slashdotters double the effect, NOW!
No such thing as a net2421 (Score:2, Informative)
Re:No such thing as a net2421 (Score:1)
Given that the person who wrote the submission is the person who wrote the project page, I think we can file this in the "accidental typo" category".
Re:No such thing as a net2421 (Score:2, Interesting)
At least I got it right on the project pages!
It'll crawl! (Score:3, Interesting)
The real speed is barely comparable to 56k modem (if the use is light on the provider's network) and this is given you have a perfect signal reception. All this is with a SINGLE computer on the network. Now if this were used amongst 2.. or more computers you'd barely be able to browse the web, much less connect to your favorite Linux box via SSH.
Re:It'll crawl! (Score:2, Informative)
Re:It'll crawl! (Score:5, Informative)
In 1xRTT-land I got 70up/30dn most of the time. About 1/4 of the time I'd get 110up/50dn. At the worst (only a few times, and usually when the evening commute hours put a lot of traffic near where I was parked) I'd get about 50/10. Compared to a 56k modem (about 26/20 on the same tool when I tried it), this isn't bad.
I've had 4 computers using it at the same time. While it will start to gronk on images with multiple access it's truly not that bad. And no, we're not using any kind of proxy, cache or compressor.
I've yet to get this system out under EVDO coverage yet, save for the single test that got me 600dn (found one local tower where it was activated. Didn't last. Hrmf). When EVDO hits my area (or I take a trip into an EVDO area) I'll put up better metrics.
Compared to GPRS (my old wireless link) it's much nicer.
Re:It'll crawl! (Score:2)
I wanna rush out and start buying parts, but I'd like to know your reasoning for that choice in particular
Units? (Score:2)
Also, your post states higher uplink speeds than downlink speeds. Is this correct? It seems contrary to most broadband connections.
Re:Units? (Score:2)
The lack of units does suck though, and it apears that he has the up/down reversed.
I use 1xRTT every now an then, and it is about like dialup.
Re:It'll crawl! (Score:2, Informative)
Re:It'll crawl! (Score:3, Insightful)
Once upon a time, before I left home, I had dialup internet. Unfortunatly my ISP was some government funded one so equipment upgrades were not really their thing.
We had a 32Kb/s internet connection. I say we because by brother and I shared this dialup connection over our home network. Now yes it was not speedy but with a good IM to keep you company it was quite useable. Image intensive pages took a little to load but we survived.
Updateing windows and downloading linux ISO
Re:It'll crawl! (Score:5, Funny)
Re:It'll crawl! (Score:2, Informative)
Note that I saw raw rate here because I'm not using the c
Re:It'll crawl! (Score:2)
Lucky HSPDA is just a upgrade to UMTS, you get devices that are backwards compatible, so don't expect more UMTS devices in the USA, all new devices are going to be HSPDA or GPRS.
http://reviews-zdnet.com.com/4520-3504_16-56649
Re:It'll crawl! (Score:2)
I'd mod you seriously misinformed, if I had mod points and there were such a mod.
gprs is barely 56k modem. evdo is not gprs, not even nearly.
the real problem w
Re:It'll crawl! (Score:2)
Seriously, it's not bad. There's a lot of latency on the network (ping times of 500 - 800 ms) so it's not ideal for ssh sessions, but I do it all the time anyway.
It's quite a bit faster than a 56k modem. And I find it works well even with a weak signal. And while Verizon may suck in many ways, they do have good network coverage.
Even better... (Score:1)
Re:Even better... (Score:1)
EVDO's coverage isn't great. The slower version (1xRTT) is about the same as Verizon's voice service. Not *everywhere* but it's been in far more places than I thought it'd be. With a 3w booster and a good antenna (a trick the long-term RV crowd has known about for years for voice use) you c
I can't even begin to tell you.... (Score:2, Funny)
Re:I can't even begin to tell you.... (Score:2, Insightful)
I know of about 5 others (two in my area) who are making similar devices... so I'm still surprised I was the first one to get pages about such a project on
How far have you gotten on yours, if you're working on the same kind of project? What else have you made yours do?
Re:I can't even begin to tell you.... (Score:2)
I'm taking notes for a project to make my entire city wireless.
Re:I can't even begin to tell you.... (Score:2)
Look at the Kenwood music keg, then look at the "About us [phatnoise.com]" link on the orignal Musc Keg's web site (aka phatnoise.com) that mentions when they first started developing it. 1999. Then search rec.audio.car for my
At least I only know of one place where someone actually beat me to market (I was gonna give my so
Car theft. (Score:1)
rural no dsl/cable option... (Score:1)
Re:rural no dsl/cable option... (Score:3, Informative)
Re:rural no dsl/cable option... (Score:2)
but there won't be nay signal out there (Score:2)
Seriously, considering how modern state-of-the-art cell phones don't work at all in such areas (at least here in the US), I don't think the chances of getting a good data link are very high.
And don't forget... (Score:5, Funny)
-matthew
Re:And don't forget... (Score:1)
Bluetooth instead of wifi gets you real mobility (Score:2)
Since its bluetooth, the cellphone stays in my pocket (unless it rings!) and the computer fits in my jacket pocket when not in use. This way I get broadband not just within range of my car, but anywhere I go.
Why do you need a car? (Score:4, Informative)
The laptop talks to the phone over the built-in bluetooth and can share the connection over the built-in WiFi card.
Re:Why do you need a car? (Score:1)
Yeah but... (Score:2)
Re:Why do you need a car? (Score:2)
Re:Why do you need a car? (Score:2)
Also check out CLIVE (Score:4, Interesting)
While you're there, be sure to check out our other hardware hack from last year, stored on the same server: CLIVE [fbrtech.com]. It's an Iridium Flare Tracker we built out of a Gameboy Advance and a DPSS laser.
I've moved all the images from both projects to the same high-bandwidth server so they shouldn't stall out any more. Being slashdotted is rather fun to watch.
Re:Also check out CLIVE (Score:2)
Glad you're having fun!
Re:Also check out CLIVE (Score:2)
I noticed the board you used has some spare IO have you thought about adding a Dallas one wire network to it? You could all sorts of sensors and digital IO to the system. Maybe a mobile weather station?
Meshing cars (Score:1)
Or... use a laptop (Score:2, Informative)
I was going to switch to an intel wifi card when the driver started improving, but they don't support master mode yet.
Under debian, it's fairly easy using ipmasq. If I "ifup" the wireless adapter when there is already a default route (from the phone or ethernet), the wifi card is set up to take a static address w
This ain't yer bluetooth phone. (Score:5, Funny)
We've had the technology and the ability to do this, but the really cool applications are just too risky, too liability-prone, or too legally questionable to catch the attention of the developers. So while your luxocar has a bluetooth network that catches viruses, and a lot of handy value-added features go by the wayside.
I mean, here's the things that are useful for an in-car network:
A) Porn. Porn drives technology, period. I strongly recommend that the next development in this field be a means to stream internet porn onto a heads-up (hands-free) display, possibly via voice command. Since we're all being open-sourcy about it, there should also be a facility to transmit and add to the global wealth of internet porn.
B) Anti-theft. This is talked about in the article, although I find it difficult to imagine a thief wanting anything as ugly as a Honda Element. Maybe if he riced it up a bit, and camouflaged the solar panel as a big-ass aftermarket wing or something; that and one of those "battery life extender" stickers that says "R-Type" on it
C) Navigation: again, there are already factory-installed and aftermarket solutions for this, but we really could use some improvements that only proper geeks can provide:
1. The author mentions networking radar detectors, as well as other traffic indicators (speed, proximity). That's a good start.
2. Much more interesting would be to network a whole slew of sensors. Radar detectors are good; but why not slap in a cheap scanner that runs through a whole range of frequencies and plots spikes and intensities? With a few sensors around, you could provide real-time plots of a large amount of radio traffic, and even localize quite a few. Heck, many police and fire frequencies are already out there on the internet.
Of course, y'all would need some centralized support for that, and if done wrong, it'd probably be the target of some congressman's ire, and attempts to shut it down.
Then again, if you ran something like a series of IRC channels (one for each region, run through port 80 and otherwise made to look like web traffic), authenticated users and blocklists, that just echoed reports from rmeote users, and maybe queries ("anyone got a picture of the tollbooth?"), you'd have your geek comms paradise, and the guy riding shotgun would have plenty of tasks to perform to isolate and avoid the mundane threats of traffic jams, separate ATIS noise from highway patrols, keep a steady stream of porn going to the driver's HU/HDD, and try to avoid throwing up.
D) Don't forget the need to bridge with existing open WiFi access points. Starbucks offers their networks as a service to the community, after all.
Then again, it's just a car. Speeding is generally something best done away from other cars. VoIP won't work too well with 3G latency. Any nerd project that gets mainstream acceptance loses most of its utility as people figure out ways to nickle and dime the life out of it.
Re:This ain't yer bluetooth phone. (Score:2)
R-Type [irem.co.jp] sticker?????
Doppler (Score:2)
-Jesse
/. effect (Score:4, Funny)
Analyzing uploaded pics? (Score:2)
Could the pics be analyzed to produce useful GIS data? At the very least could street names and circulation signs (no left turn, etc...) be recognized on pics and geocoded?
There's a lot of interesting social software that could be built with even partial data, but the cost of the commercial sets is so high that it discourages casual hacking.
Can this be done on a GSM network? (Score:2)
I see that the Verizon card is CDMA. Can you build a StompBox that uses GSM networks? I'm assuming you have to find the right PCMCIA card [ebay.com] and just swap out your Verizon card and recompile your kernel. If it could be moved from 1xRTT to GSM, it could be portable to just about any GSM network worldwide! Imagine taking the box around Europe and still have internet access to email home!
Nice! (Score:3, Insightful)
Re:Nice! (Score:2)
Using google maps with GPS? (Score:2)
Re:Great... but (Score:2)