Generic internet appliances

Submitted by brad on Wed, 2006-12-13 00:54
Topic: 
Internet

Normally I'm a general-purpose computing guy. I like that the computer that runs my TV with MythTV is a general purpose computer that does far more than a Tivo ever would. My main computer is normally on and ready for me to do a thousand things.

But there is value in specialty internet appliances, especially ones that can be very low power and small. But it doesn't make sense to have a ton of those either.

I propose a generic internet appliance box. It would be based on the same small single-board computers which run linux that you find in the typical home router and many other small network appliances. It would ideally be so useful that it would be sold in vast quantities, either in its generic form or with minor repurposings.

Here's what would be in level 1 of the box:

  • A small, single-board linux computer with low power processor such as the ARM
  • Similar RAM and flash to today's small boxes, enough to run a modest linux.
  • WiFi radio, usually to be a client -- but presumably adaptable to make access points (in which case you need ethernet ports, so perhaps not.)
  • USB port
  • Infrared port for remote control or IR keyboard (optionally a USB add-on)

Optional features would include:

  • Audio output with low-fi speaker
  • Small LCD panel
  • DVI output for flat panel display
  • 3 or 4 buttons arranged next to the LCD panel

The USB port on the basic unit provides a handy way to configure the box. On a full PC, write a thumb-drive with the needed configuration (in particular WiFi encryption keys) and then move the thumb drive to the unit. Thumb drives can also provide a complete filesystem, software or can contain photo slide shows in the version with the video output. Thumb drives could in fact contain entire applications, so you insert one and it copies the app to the box's flash to give it a personality.

Here are some useful applications:

  • In many towns, you can see when a bus or train will arrive at your stop over the internet. Program the appliance with your stop and how long it takes to walk there after a warning. Press a button when you want to leave, and the box announces over the speaker a countdown of when to go to meet the transit perfectly.
  • Email notifier
  • MP3 output to stereo or digital speakers
  • File server (USB connect to external drives -- may require full ethernet.)
  • VOIP phone system speakerphone/ringer/announcer
  • Printer server for USB printers
  • Household controller interface (X10, thermostat control, etc.)

Slap on the back of cheap flat panel display mounted on the wall, connected with video cable. Now offer a vast array of applications such as:

  • Slide show
  • Security video (low-res unless there is an mpeg decoder in the box.)
  • Weather/News/Traffic updates
  • With an infrared keyboard, be a complete terminal to other computer apps and a minimal web browser.

There are many more applications people can dream up. The idea is that one cheap box can do all these things, and since it could be made in serious quantities, it could end up cheaper than the slightly more specialized boxes, which themselves retail for well under $50 today. Indeed today's USB printer servers turn out to be pretty close to this box.

The goal is to get these out and let people dream up the applications.

Share on:
  • Reddit logo
  • Facebook logo
  • Twitter logo
  • LinkedIn logo

Comments

David Orban

Wed, 2006-12-13 02:17

There is a company, and a product platform that you should look at: Chumby, http://www.chumby.com. They have a great team, a wonderful vision of open source hardware complemented by valuable services. The first products will be shipping soon, and they have developer programs in place now.

David

Ken Kennedy

Wed, 2006-12-13 10:38

Hey Brad...I second the Chumby comments. It's going to be a little more expensive ($150 range), and I agree that $50 would be awesome. But I'm really looking forward to it.

Also, it's more a geek solution, but you can do a lot of this by repurposing many routers (WRT54G is a good example, though no USB), as I'm sure you're aware. Openwrt.org is a good example. I agree that USB is a great way to configure and update these sorts of devices; it's quite annoying that USB connectors are so rare on them.

Your point, of course, is to not have to do all the geek hackery, and I agree 100%. But it's a start.

rjh

Wed, 2006-12-13 10:54

You could also look at the WRAP PC-Engine. It comes close although it misses on a few points. I have one that is my customized firewall/proxy machine.

brad

Wed, 2006-12-13 18:43

Realize that designed as I propose, companies like Linksys and the rest would find it cheaper to build their routers, print servers, file servers and the like based on the generic internet appliance, either as-is, or with minor mods. They could then take advantage of the large quantities they would be made in, and on competitive bids from companies making the boxes.

$150 isn't an appliance, that buys you a full geode computer these days!
Yet cheaper routers (which are this box plus access point radios and antenna, and without the USB) typically retail now down to $30.

rjh

Thu, 2006-12-14 09:33

I'm not sure that this will work out as you expect. I learned some interesting things while handling CAD/CAM for a large corporate with an industrial products division and a consumer products division. The engineering and design cultures are much more different than I expected. An engineer shifting from one to the other would take a year or two before the automatic reflex judgements would adapt.

At the core of this is the simple cost tradeoff related to volume. Start by using a nominal cost of $250K/yr for an engineer, lab, prototypes, etc. A consumer product will have a production run of say 2.5 million. So it is worth a year of development engineering to save ten cents. A dollar of savings is worth ten man-years. Industrial products have typical runs of 25,000 making it take a $10 cost savings to justify a man-year of engineering.

It sounds simple, but the ramifications reach everywhere. One of them is that the industrial engineers automatically think "standard part means good, saves money" while the consumer engineers automatically think "standard part means bad, costs money". Finding and understanding all these differences is part of why it took so long for an engineer to shift divisions. Both the industrial and consumer engineers are right in their view of standard parts. Standard parts reduce the up front engineering cost, but they carry the price tag of a slightly higher production cost for unused features.

Your appliance idea is drifting into the transition area where sometimes custom designs will be cheaper and sometimes standard components will be cheaper.

You are right that $150 gets you a full geode computer. My WRAP machine, enclosure, power supply, and CF card cost about that for a retail purchase. It is a geode computer. In volume the pricing is more like $100. This is cheap enough for some interesting uses, not cheap enough for others. My actual motivation was eliminating space, noise, and electricity consumption. Instead of an old PC with fans, etc., I have a roughly DVD sized little box. I have an old Linksys that I considered adapting, but I want the combined function of firewall, multiple application proxies, and traffic logging. It will not all fit into the 8MB Flash of my Linksys box. It takes about 32MB for all that.

If you disassemble the Linksys routers (there is a web site somewhere by someone who did just that) you see that they are partially into the consumer design territory. Their "linux" model retails for $60 while their custom model retails for $40. The difference is a lot more custom components and custom design for the $40 model. It has the bare minimum RAM/ROM, it has a more highly integrated and specialized ethernet circuitry, it has a custom integrated wireless, it has eliminated most I/O options.

As a result, it is immensely harder to modify for other purposes or even to extend with feature additions. They cut the RAM/ROM so far that even highly stripped Linux kernels have trouble fitting. Their older systems had enough room for adaptation and modification. These new ones are so much harder to use that customers will find it easier to pay the extra money for the "linux" version.

This industrial volume versus consumer volume difference is going to be significant for the internet appliances.

brad

Sat, 2006-12-16 13:14

Actually, one of the big differences of the new model of linksys is it has less memory on it. They decided, as you say, that a few bucks costs a lot over millions of units, and wanted to make a version that didn't need all the memory a linux unit needed.

I agree a consumer product with a very large market could underprice the generic appliance I describe. So if you are going to sell a 3 million home gateway boxes, it will be better to do it custom. But for the up and coming entrepreneur with a cool new product idea who will only sell 200,000, not because the product isn't as good or better than the big market product, but because the small company doesn't have the marketing muscle, a generic base would indeed be just what you need.

Their version 2 product, after they become a big success, might well be custom hardware aimed exactly at their product spec. That's fine by me.

Add new comment