Hacking the Actiontec 56k Modem/Gateway

william_lorenz writes "The Actiontec Dual 56k External Modem is an inexpensive device with a built-in 56k modem and two Ethernet ports that can be used as an Internet gateway of sorts. What's great about it is that it runs some form of uClinux, it's easily hackable, and Greg Boehnlein of the Linux Users Group of Cleveland and NOOSS fame recently contributed a detailed report on his findings! Pictures of the board are also available here, here, and here. Lots of specific details are included in Greg's article, and there's been some further discussions about this on the LUGC mailing lists."

Comments

Let the comments begin!

[Anonymous Coward]

Jeez...

Slashdotted already.. this article's gonna set the record for redundant posts.

Yes, we KNOW his server MUST be behind that 56k modem.

Right, now that we've got that out of the way...

Hmm... other products

[Empiric]

Personally, I'm more intrigued by the company's anti-kidnapping technology. I'll sleep easier once that's out of the way.

Sponsored by AT&T

[madsen]

From the ActionTec site:

the External Dual PC Modem keeps you connected longer, and faster.

Re:56k gateways

[mackstann]

Two ethernet ports + linux + easily hackable = who cares about the modem jack?

They say it's hackable

[Anonymous Coward]

"it's easily hackable"

If only we could hack it into a 256k modem...

Re:They say it's hackable

[Anonymous Coward]

It's not really the FCC's fault. They imposed the limitation due to the potential for crosstalk due to the nature of the lines themselves. That said, specifying it in bits/second was a dumb way to go about it (and no doubt suggested by the telcos) -- at least, I think that's how the rule is written. So, back in the days when switching was still physical, it would be likely that, were everyone using these hypothetical 256k modems, you wouldn't be connecting near the full speed due to all the interference, and voice service would be degraded for everyone.

But in any case, *since* the regulation was put in place, this allowed the telcos to move to digital switching, since they could know the limit of the signals they'd have to digitize to support all legal users/devices on their circuits. You have to admit, it would've been hard for them to carry on any other way -- Imagine if they still had to support racks of relays just to ensure your attempt to run ethernet-over-phoneline would work. Now imagine if they had to try to support that for *long distance* calling...

Now, DSL is a little different. For one thing, it terminates at the telco's central office racks, or a 'remote terminal' (basically a central-office-in-a-box that connects back to the main office over fiber or some extreme-speed-over-copper solution), so the telco only has to support line quality over a limited, known distance. For another, it takes advantage of technological advancements to run up in relatively high, definitely inaudible frequencies*, so there's little risk of corruption to others' voice service... and further, the hardware is, in fact, advanced enough to pick from any of a number of 'channels,' such that, while it's still only a point-to-point tech, it can avoid crosstalk from your neighbor's link by running on completely different frequencies. Practically speaking, it *is* the best thing telcos can bring you right now over existing copper && at a vaguely affordable price && while being profitable enough for them to continue deploying it.

Now, in exchange for this non-switchability (you aren't "making calls" with DSL, it's basically like having a big fault-tolerant null-modem cable between you and the terminating equipment), the telcos are supposed to open up their racks at reasonable cost so other ISPs can come in and give you a choice of whose service you're plugged into. Yes, this part has been getting a bit screwed up, but that's a political issue; it's hard to imagine it working another way technologically... unless they'd done something like electricity 'deregulation,' and set it up so you'd *always* be using the telco's equipment, backbone, netblock, etc., and choosing which ISP to pay the equivalent of your 'generation charge' to. If you think about it for a moment, you'll realize that'd work roughly as "well" for telecom as it has for electricity, so perhaps it's a good thing they didn't try it that way.

So the real question is not, "Why do we have to buy DSL instead of 256K modems?," it's "Why do we still bother with 'voice' circuits at all, when everything could be 100% digital and routable, with Vonage-style boxes at the NID of the homes?"
It's 2003, and it'd make the most sense for the telcos to become 'data utilities,' in competition with cable, wireless, and third-party fiber-stringers... ...but we're still in a transition period, and since the telcos *will* leap like rabid weasels to price-fix and destroy competition if they're let off their leashes, it's not clear what we can do. The same regulations that *allow* competition for DSL (forcing 'open racks') make it unattractive for the telcos to offer DSL without the underlying, obsolete voice circuit - because then the ILEC wouldn't be getting their cut (or at least, as big of one) if you opted for a CLEC's service.

Meanwhile, of course, cable is practically unregulated, so you get your local monopoly with that, and absolutely no ISP choice (which would be technically 'expensive' anyway, giv

possible to hack cable/adsl routers?

[narkotix]

Would it be possible to hack an adsl/cable router to be used as a simple webserver? For a low traffic and static site it would be perfect for my business website hosting needs. 10watts consumption...fanless operation...and small footprint it sounds like a dream! I could even imagine other hacks like interfacing a larger amount of flash storage or running a real low end db &scripting engine to have behaviour like mysql/php3 together.

Re:possible to hack cable/adsl routers?

[zakezuke]

So it's a cute idea, probably do-able, but not worth it in practice. You'd be better off going with a cheap bare bones system running Linux or FreeBSD. No CD or floppy drive and underclock it to reduce power consumption.

One thing I know i've discussed over a pint of beer are remote observation stations, trivial little devices that measure temprature, water level, that sorta thing. Out of the way places with NO easy access to landlines.

If you are talking off the shelf barebones system, you are talking a minium of 60watts for basic option. 10watts is a hell of alot more attractive if your power source is something like solar and battery storage.

Re:Is the page being served by that modem?

[amcguinn]

What next? RAID-5 using a stack of 8 inch floppies?

Too late

Re:56k gateways

[zakezuke]

56k is better then nothing. Keep in mind also the fact that primarly text information gets compressed unlike in broadband technologies. Keep in mind that many a small business or small remote office doesn't nessicarly need lots and lots of bandwidth just to get e-mail.

Hackability? Well I'm somewhat curious what they can do with such a device. The first thing that comes to mind is a standby gateway that goes online when the primary gateway fails. This would be MOST handy.

NOT "dual modem" -- rather 1 modem, 2 users

[sonamchauhan]

The "dual modem" mention in the article header made me think for a moment this product allowed multiple users to share a composite link. (See my earlier post on this topic). Rather this product allows 2 users to share *one* link.

A composite link to two *different* ISP could be implemented quite simply by say, using a proxy server to multiplex outbound HTTP requests among multiple interfaces (each interface corresponding to one phone connection).

This approach is more coarse-grained and inefficient than TCP/IP-level channel bonding. However, it would still be useful for places out in the boondocks where you can get two telephone lines, but no broadband. Also, its efficiency could be improved by using HTTP functionality that allows specific byte-ranges to be downloaded for a particular resource.

Re:Who says modem must be used outgoing ?

[pe1rxq]

The only downside of your setup is that the 56k modem drops to 33k6 when you use it to dial in...
56k only works if the isp can tap into the digital phone exchange...

Jeroen

DirecPC server?

[crypton]

Sounds like this might be usable for anyone wanting to set up a server for a satellite connection that has the downlink via the dish/usb modem and the uplink via ISP/phone modem like DirecPC does. The only linux project so far is at Sourceforge: http://sourceforge.net/projects/direcpc

Dammit, where can I get one?!?!

[Guppy06]

Laugh if you want, but necessity dictates that I can only get a 56k internet connection, and there's two computers in the house. I'm currently using XP's internet connection sharing thingie but I'd kill for a hardware-based solution right about now, at least one that doesn't cost an arm and a leg. I thought Stratitec's (discontinued) "Easy Internet Router," which had a serial port for connecting an external modem as well as ethernet ports, would be The Answer, except that the router seriously slowed down when routing a dial-up connection (tech support's excuse was "Well, we only intended it to be a back-up and the router really isn't designed to do that...")

Now I hear about this, which to me sounds like a Holy Grail, and I can't seem to find it anywhere. The only place I can find it is one of those shady dealers operating on Amazon. There has to be somebody slightly more reputable with some for sale, or at the very least something else to let me compare prices!

Samping

[Jennifer E. Elaan]

I was going to mod you up, but I decided to reply instead.

The limitation you refer to has very little to do with managing crosstalk. What actually happened is that the phone company was maximizing the number of channels they could fit on a single wire by using a technology called Frequency Division Multiplexing (FDM).

Since an average voice conversation has a bandwidth of about 2KHz, they built in a low-pass filter with a cutoff somewhere in the vicinity of 3KHz. This means they can heterodyne the channels, each (roughly)3KHz wide, onto a single wire.

Now, this means that the data rate (in terms of zero crossings per second (the original meaning of baud) is limited to about 2400. The "high speed" modems, all the way up to 56K, have a baud rate of 2400. This is a hard limit due to the phone company hardware.

What changed is the number of bits per baud. A 56K modem might use as many as 24 bits per baud, assuming the line is clear enough. The number of bits per baud is capped by the noise floor of the signal, which is also why you won't always connect at 56K (noisier lines can't handle the resolution).

In the move to digital networks, the same total channel datarate was designed into the switching systems. I'm not entirely sure the sampling rate and quantization parameters of these systems, though.