Slashdot Mirror
DSLBlaster?
The Jake writes: "Here's some mindless link propagation for you, Kuro5hin style. Memepool reports that AuDSL will let you use an old sound card to connect to a subscriber line. "AuDSL is an experimental technology for achieving low-cost leased-line Internet connectivity for homes and small businesses. The acronym AuDSL stands for Audio Digital Subscriber Line. The idea is to replace traditional, expensive leased line modems with software modems running on PCs, connecting the leased line to an ordinary PC sound card. This makes it possible to construct a complete leased line internet gateway entirely from inexpensive commodity PC hardware.""
True, though I would think that this is a good thing to have worked on, if only to provide a short-term fix when appropriate hardware isn't readily available. If you need your network connection, but if you can't get a new DSL or leased-line modem for several days, this would be a good thing to try.
And no, I don't think it'd be a good idea to play quake on a machine that's doing this. It's something to be used on a router, IMHO.
I had been curious about this sort of technology a year or so ago, but my math must have been off, and I figured it wouldn't work any better than a standard modem. I'm glad to see that other people had the same idea.
--
Pronounced "kuh roh five heen," kuro5hin is a weblog for the l33t. After all, if slashdot was l33t, it'd be named Slashd0t. Kuro5hin did not make a similar mistake by picking something inane like "kuroshin" as its name.
10 PRINT CHR$(205.5+RND(1)); : GOTO 10
I've read the damn FAQ. Hell, I spend more time there than here. It's still pronounced "kuh roh five hin," and it's still l33t. More l33t than you anyway.
10 PRINT CHR$(205.5+RND(1)); : GOTO 10
So does this mean we'll get a better connection if we use monster cable, just like the guy in the audio store tells us? :)
consult the Jargon Dictionary ... one link is:
_ re incarnation.html
...
http://info.astrian.net/jargon/terms/w/wheel_of
The basic premise is that as CPUs become more powerful, that CPU assumes more of the tasks once relegated to special-purpose hardware. Then, somebody else notices that the CPU is doing "low-value" work, so the tasks move into dedicated silicon
Of course you completely ignored the various compression techniques in use today. Also you are assuming that 48KHZ is a 1 bit mono signal. However soundcards typically have a datarate of 48KHZ * 16-bits * 2 channels = 1.5Mbit / second . That comes pretty close to basic DSL. Taking into account modulation / error correction, that should still leave a sizable amount of IP bandwidth available...
-adnans
"In short: just say NO TO DRUGS, and maybe you won't end up like the Hurd people." --Linus Torvalds
The idea behind DSL was to eliminate the usage of acoustic signals for data transfer. By modulating the DSL signal into an audio stream, you're defeating the purpose.
Sound cards aren't even designed for this sort of thing. The same limitations that hit Winmodems will apply here as well, and will be multiplied. There's going to be some serious CPU usage (which could eventually be partially alleviated by the DSP development) because you still have to modulate/demodulate the DSL signal from audio to data. Not only that, but a specialized circuit must be built to even do this.
I would assume the low cost wouldn't be a factor due to the hassle, and the functionality.
This type of connection will yield slightly-faster-than-Dial-Up speeds at DSL prices. How does that constitute "Low Cost"?
-- Give him Head? Be a Beacon?
-- Give him Head? Be a Beacon? :P)
(If you can't figure out how to E-Mail me, Don't.
Let's see... A typical stereo sound card has 2 A/D-D/A channels sampling at 44 kHz at 16 bits per sample. So that gives us 44,000*16*2 = 1.4 Mbps. Not too bad I think, but most likely in practice you'll get only a small fraction of this, quite possibly comparable to 128k ISDN or less on the average. A SB16 can only do full duplex with either the A/D running at 16 bits/sample and the D/A running at 8 bits/sample and vice-versa. And one of them will have to be restricted to mono. Which means maximum upload bandwidth will probably be less than a quarter of the download bandwidth, which is not too bad as for the average user, this is made up primarily of HTTP/GET requests :).
Qu'on me donne six lignes écrites de la main du plus honnête homme, j'y trouverai de quoi le faire pendre.
I've always been very skeptical of the claims surrounding the notion of transmitting data at high speed over power lines.
Even if power line data transmission can do all they claim it can (which is doubtful given the reasons you've listed), there is still one fundamental problem to be overcome.
The bandwidth will be shared. If it becomes popular, you're back in the same boat as a cable modem (and to a lesser extent DSL)... a slow connection with high latency.
Bah. If they spent their money on research to make fiber optic lines cheaper to install, they'll be a lot more successful and would make more money.
I mean, why don't they figure out how to trench in a cable more quickly and cheaply? Or how about some radar detection system that would warn you before you dug into a buried power line or water pipe? Or a badger-type robot that digs a hole and drags a cable behind it?
Hey, I already thought of this... sorta. I was thinking that you could hook two computers up over a phone line by using speech synthesis/recognition software. Think how much easier it would be to debug communications issues if you could understand what the computers were saying to each other by eavesdropping...
computer 1: EA BF 02 51
computer 2: Huh?
computer 1: I *said* EA BF 02 51 36
computer 2: Oh, yeah.
You're not supposed to ask why. Just chuckle and appreciate the Hack Value.
---
As copyright owner of this comment, I authorize everyone to defeat any technological measure which limits access to it.
Reminds me of an old, crazy idea I had once...which would be to do a bus-topology network using sound cards.
As sound cards don't have some sort of GUID like an ethernet hwaddr, one channel of the stereo sound cable would be used for constantly broadcasting the node's address, and the other used for data.
Was never serious about it, but it was funny to think of...and now someone's actually DOING networking with audio cards!
-- Veni, vidi, dormivi
Wonder what would happen if one accidentally piped a BS MP3 through the sound blaster while it was connected to the line? Would the perky umm.. lyrics boost the line speed? :)
For that matter, why not use ethernet?
I hear they've got ethernet cards which work at gigabit speeds. That's like, 100 megs a second!
Even better, I hear the PCI bus is good for 132 megs a second. Let's just build a PCI to PCI bridge, with 10 miles of telco copper in the middle!
It's digital. It must be better, right?
Nevermind that the telco line won't carry the signal. Nevermind that it wouldn't carry SP/DIF, either. The shit is made for analog signals. It's -good- at carrying analog signals. Thus, analog signals are the order of the day.
Gee, back to the drawing board.
(think first, post later. k?)
Kid-proof tablet..
Sure they put DSL in the name, but they operate at 96kbps, or a bit less than twice modem speed wire. My little DSL line is pluggin along at 12 times that rate.
That 96kbps is over a spool of wire. You can not get more than 64kbps through any POTS line that goes through the 8KHz, 8bit conversion in a digital telephone switching system. (If you think you can, then I suggest you find a more lucrative use for breaking the laws of mathematics.)
Further, they make no mention of cross talk sensitivity or generation. Remember that in the US our modems are only 53kbps anyway because of frequency limitations.
I give it an A for inspiration and a D for utility. The FCC will likely give it an F for violating power spectrum limitations on a phone line. (Putting DSL in the name was pure marketing genius! AuModem would never have gotten on slashdot)
Yes yes, I know. Its not funny. But I'm bored so I had to post SOMETHING. :)
-Restil
Play with my webcams and lights here
For Sale: 10 aging "DSL modems" in my closet SoundBlaster brand...
Look at your units. You're multiplying bits by Hz (which are 1/seconds). That gets you bits/second, which is, incidentally, what bandwidth is.
Perhaps this is why Physics tends to be a required course for Comp Sci majors...
--
Win dain a lotica, en vai tu ri silota
If things are going to be done this way, couldn't they at LEAST use a PCI sound card?
"Everything you know is wrong. (And stupid.)"
"Everything you know is wrong. (And stupid.)"
Moderation Totals: Wrong=2, Stupid=3, Total=5.
Although it is true that some things can be implemented cheaper in software, this is not necessarily true all the time. (And of course cheaper != better)
Is our future generic appliances that can upon downloading the latest software release do whatever we please. Although this would be nice, it is hardly realistic. It flies in the face of years and years of solid task-appropriate engineering.
Furthermore, I don't know anyone who doesn't have DSL because of the price of the damn DSL modem. Its like not having nike's because you can't afford to be replacing the shoe laces.
Software is great for implementing more niche-market type appliances (remember firewalls in the early days.) But the limitations and unnecessary over head of generic solutions will eventually give way to well-engineered solution-specific devices.
I thought winmodems were a thing of the past, but it seems someone is trying to reincarnate them in the DSL flavor. I am sure they will find the limitations of the sound cards a huge-tradeoff for the supposed money savings.
I mean seriously, what they are doing is using their souncard for a tradition modem.
doh!
Sigs are awesome huh?
The cards were designed for audio - they have filters which remove aliasing artefacts from the output. Sounds better, but it limits the high-frequency response (the filters don't have infinitely sharp cutoff) so the effective dynamic range is less than 16 bits. Also, the Shannon limit is only half the sampling rate, so even without filters you'd "only" be able to get 384 kb/s.
--
So... could I get 6 times the speed using the 5.1 channels in via SPDIF that my SBLive supports?
Better yet, hook up a line to one of the inputs on my stereo. Leave it on as white noise and figure out when the best time to download porn is by how quiet the other traffic is.
In place of my cable modem, I'll hook the cable into my television set. It'll show a scribbly image, but the TV's radiation will minutely interfere with my fluorescent light. By reprogramming the DSP inside my optical mouse and holding it up to the bulb, I'll have a steady stream of data coming through the mouse's USB connector. From there, it's a simple matter of hooking the mouse to my Laserjet's USB port, changing the ROM to redirect the packets out the parallel input port, running that to my Xircom parallel port ethernet adapter, and from there straight into my 10 base-T hub.
Now if I can just think of a way to get from the hub to my computer....
Go ahead, moderate this a troll. I can take it.
But seriously, why post this story? It's an utterly useless hardware hack which is tremendously behind the state of the art and (IMHO) doesn't even qualify as particularly creative.
Using audio signals to transmit digital data!! Woohoo!! That's never been done before!! Now, the flourescent light network story was pretty interesting, and at least relatively novel, but this is pure drivel.
If you read the tech specs, they're only getting 96Kb out of this thing, and it requires a leased line. Going through this amount of work to get twice the performance of an ordinary modem seems a bit ludicrous. What I'm amazed about is that they're only getting 96Kb -- I'm no signal expert, but that seems positively anemic given a 48KHz sampling rate. They're only getting two bits per sample?? WTF?? Truly pathetic next to my 1.5Mb DSL at home.
This is news because... why, exactly?
ZFS: because love is never having to say fsck
As it says at their site, it is currently only able to reach speeds of 96 kbps. Why not use a modem instead?
Other things worth noting: "Running on a 333 MHz AMD K6-2 processor, the software modem consumes about 38% of the CPU cycles." - Ouch. And that's on Linux. I wouldn't want to try something of the sort on Windoze with a few browser windows and an app or two open.
Could this technology become interesting? Perhaps... maybe in three years when they have the bugs worked out and the code optimized, you can use it to get 256 kbps, maybe... Personally, I'm not planning on waiting that long!
There are so many better solutions out there, and if you want my opinion, the real future of broadband is in the wire that is already running into your house. No, not the phone wire. This one. Now serving over 90% of the world's population...
You can raise a kid with one parent but it shouldnt be done
You can drive a car with your feet but it shouldnt be done
You can fertalize you lawn with used motor oil but it shouldnt be done
And now you can build a DSL modem out of a soundcard but it shouldnt be done
A couple of Speakers and mics, and you could have a cheap wireless network!(though, the noise may drive you a little nuts).
They ought to be able to do much better given a solid copper pair, 16-bit D/A and A/D, and a 48KHz sampling rate. The Shannon limit for that setup is 48000*16 bits/sec, or 768 Kb/s. That's under optimal conditions, but ordinary modems routinely get 80% of the Shannon limit for their medium (8 bits x 8KHz sampling rate = 64Kb/sec) The line has to be characterized during startup and appropriate equalization applied, but that's a well-understood process built into every modem.
ADSL modems (at least; don't know about SDSL) aren't actually that expensive. They can be had (used) for $50-$100 if you do a little looking. They're aren't very complicated devices. The reason they cost alot is that the manufacturers currently are only supposed to sell to DSL providers: the end user can't just pick one up at CompukeSA, so the providers charge what they want.
So what I don't understand is, why would the providers ever go for something like this?
Oh yeah, and don't even think about netgaming on this system. Soundblasters have a bad habit of taking over the PCI bus as it is. I can't imagine this making it anything but worse.
"Research is what I am doing when I don't know what I am doing." -- Wernher von Braun
Why can't you use wireless ethernet?
If it ain't broke, it doesn't have enough features yet.
A DSL line gets its speed mostly by virtue of not having to fit the bits through a 4khz filter. Shannon tells us that 33.6 is about the limit for a 4kz signal - 56K modems actually cheat (hint - *every* computer store stocks the "consumer" end of a 56K modem - ever tried to buy the *uplink* end of the pair, and provisioned the line from the telco? ;) Also, there's no pesky power restrictions - those enter into it because the power determines just how close to a perfect square wave youy can get (since the power is basically the slope of the near-vertical parts of the waveform).
A while back, I attempted to scrape together a wireless ethernet connection of SOME type out of some modems and a couple of cheap-O walkie Talkies...
Needless to say, it was a TOTAL flop. The walkie talkie transfer was garbled beyond recognition and I couldn't even connect the two points at 300bps. (I seriously have to question whether it would have worked at all to start with though- I'm not a genius with funky software hacks, but I know enough to hang myself with...)
The POINT was to build a better way to transfer MP3's to the hard drive in my car's mp3 player--yah- one more keen ornamental hack, but it would be much nicer to do it that way than physically either lug the whole unit in or string a frikkin LONG cable all the way out to the garage...
So I'm thinking this sound based hack PLUS a couple of those high tech mini-fm 2 way radios that seem to be everywhere now, and this just might work!
Anyone else been down this road?
Sig currently under construction. Mind the gap....
No kidding. You forgot noise.
Just because you slap 16 bit D/A/D gear on the ends of the wire doesn't mean you're going to get 16 bits per symbol through it, sweetie. Hell, why not use 32 bits? 128? The sky's the limit!
Shannon also says that you have to divide in the noise level to figure out how many bits you can get per symbol. THEN you design your D/A/D gear to suit. Dialup PC analog phone modems use 2400 symbols per sec x 14 bits per symbol for 33.6 kbps -- 2400 Hz because of the phone system audio sampling rate and 14 bits/symbols because of the expected noise floor and maximum allowed signal power (FCC limit). 56 kbps cheats the noise floor down a bit but I won't go into that.
Now, *IF* you've got a LEASED LINE that you can dump massive signal power onto (i.e. no other equipment on line), and you've got low inter-symbol interference (ISI - think of it as a time smear of the symbols), then hell yeah you can get 768 kbps. And I'd drive 130 MPH to work if it wasn't for those pesky cops (and curves).
One simple rule for its versus it's
Where does Kuro5hin.org get its name? (Go To Top) No, it is most decidedly not 'l337. It is a pun on Rusty's name. Kuro5hin == corrosion == rust == rusty.
look here---> Kuro5hin origin
Don't eat shrimp candy, just a heads up.
My theory is that they are so snowed in for most of the year they sit in front of PC's and code. This is just cool.
--
Jon - TheSpork
. . .but I'm not entirely convinced of it's feasability.
Let me explain.
Using audio to transfer data isn't a new idea. The humble telephone line modem bears witness to that, as does modern DSL.
What is new about this is the really creative hardware hack. Sound cards (like any other expansion card) are manufactured with a purpose in mind (in this case, that purpose is producing or inputting sound). With the exception of some cards produced with a modem input, these adapters may not be designed in such a way to enable data transfer.
That would have to be addressed in software. To make this method widely-used, AuDSL would need to include modified drivers for common sound cards in addition to the software instructions that tell your computer how to take this audio data and turn it into something usable.
And then there's the matter of it taking 38% of CPU cycles. . .
A better solution might be to design a less expensive alternative to already existing hardware that would work with already existing systems.
I admire their ingenuity, but I wonder if there isn't a better way. . .
So all we need to do is encourage this trend until we get to the point where someone figures out a way to simulate the processor with software and Presto! computers cost nothing to replicate, generate no heat, and can be shared over Napster.
-- MarkusQ
Next week: How to legally avoid paying taxes without having to earn any money.