« Content Watch Quality Assurance Manager | Main | Don't Send Word Documents to Norway »

T-Engine: Ubiquitous Computing

Back when I started computing, I used Fortran on an IBM 370 and timesharing via an IBM Selectric-like teletype. Many users—one computer.

At that same time, I got my first taste of microcomputers, what we’d call personal computers now, by building a MITS Altair with an 8080 CPU and 4K of RAM for the College of Mines (my undegrad is in Metallurgical Engineering). One user—one computer.

Now, of course, I routinely carry 4 or 5 computers around with me all the time (if you count things like my iPod and Canon S500) and I’m dependent on computers being everywhere. One user—many computers.

The problem of course, is that these “ubiquitous” computers cooperate imperfectly—some don’t cooperate at all. For example, the computers in my pick-up and the computers in my laptop bag don’t know the other exists. Even hooking the computer in my iPod to the computer in my car stereo, a no-brainer from a functionality standpoint, requires interfacial handstands.

Now, a Japanese-government sponsored research consortium that include five chip makers and 17 other Japanese high-tech firms, has announced that the T-Engine, a ubiquitous computing platform is ready for prime time. T-Engine is a platform that developers can use to build ubiquitous computing solutions from commercially available products. The platform includes four different types of T-Engine boards:

  • Standard T-Engine are used for smart phones and other portable information gadgets;
  • Micro T-Engine are used for devices with relatively basic user interface functions like a toaster or microwave oven;
  • Nano T-Engine are used for for small home electronic appliances like a stereo receiver; and
  • Pico T-Engine for adding ubiquitous computing to the smallest of components such as valves, sensors, switches, and so forth.

The platform runs a real-time OS called iTron. Toyota is using T-Engine components at a smart house demonstration at the Nagoya Expo.

T-Engine includes built in support for responding to RFID chips, bar codes, smart tags, and so on. This provides for some interesting uses.

Japan’s Ministry of Land, Infrastructure, and Transport is spearheading the testing of the UID Center’s Autonomous Movement Support Project. Electronic tags embedded in pavement stones and street furniture will supply users with location-specific information “anytime, anywhere, to anyone.” In the cities of Kobe and Tsuwano, hundreds of electronic tags and road sensors have been embedded in the pavement, sidewalks, and street furniture, providing information to tourists about historical sites and to wheelchair users about obstacles.

You could imagine a world where everything is tagged and screaming out its location to anything that asks. There are, naturally, all kinds of privacy and security implications, but I’ve found that that’s the definition of interesting problems: if there aren’t security and privacy implications, you’re probably not doing anything very fun.

You may be thinking that a project of this scope doesn’t stand a chance of becoming the dominant embedded computing system, but I think there’s at least a good chance. After all, Japan makes a great number of the cars, electronics, and appliances that are consumed in the US and the rest of the world. What’s more, the Japanese government has shown their ability before to orchestrate entire industries for national advantage. With any luck your 2007 Toyota Camry and your Mitsubishi food processor will be exchanging recipes in the not too distant future.

Posted by windley on June 27, 2005 1:13 PM

See related posts:

6 Comments

Comment from Mark Hahn at June 28, 2005 7:26 AM

ubituitous will never matter until there is a rock-solid, auth/encryption infrastructure in use by even the smallest tags.

Comment from JohnO at June 28, 2005 7:42 AM

Most users, for basic things (*not* online banking for example) don't regard privacy as highly as you do. Therefore it won't matter that their neighbors know they have a toaster in their house...

I remember similar claims for iTron a long time ago (was it as much as a decade or two ago?) And of course there was the Japanese "5th Generation" computer project that was supposed to take over the world.

Is it open source? If not it still won't go anywyhere very far.

This should not be a chip architecture. It must be an open API architecture with open source reference desgins to go anywhere.

There is no reason that ARM or other processors couldn't do this and different processors will want to be in different niches based on the volume pricing and capabilities. Silly to force to one processor architecture in this day and age.

Comment from To Mark Hahn at June 29, 2005 11:55 AM

Yes, most users won't care if their neighbors know they have a toaster. Until their neighbors know that they have, a toaster, a fridge, what's in the fridge, a large plasma tv, $20k in stereo equipment, and a small safe. This information is also available to the guy in the moving van pulling up into the driveway, but of course, nobody is moving.

Hi, actually security is built into the system architecture. A piece of hardware called eTRON is used for encyption, authorization, securing communications, and for transactions like ticketing, etc.

The platform is based on open specs that are created through a community process. Also, the software is going to be open source - though only some parts are open at this time. But, the software is not supplied under GPL, but under a different license, that's more suitable for embedded systems.

However, it's not targetted at being a completely open source system (in that others are not asked to open their sources), since hardware is becoming a commodity and software is some times the only thing that people can sell...

There's much more to the Ubiquitous Computing scenario and I think this is a good start.

Cheers,
Mohit.

p.s. Shameless plug - I maintain a web-site for the T-Engine in English. Hopefully, it will become a central resource for developers in English - http://www.onghu.com/te/ Right now, there are some fairly useful (IMHO) links there.

Comment from zephyrus at July 8, 2005 9:52 PM

-- JohnO wondered >Is it open source? If not it still won't go anywyhere
very far. Also he said, "This should not be a chip architecture. It must be an
open API architecture with open source reference desgins to go anywhere."
.........br
.br

The offerings from the TRON project cover both bases.

First of all, ITRON itself is a "specification" for RTOS. It
is open and available, for example, at:
http://www.sakamura-lab.org/TRON/ITRON/SPEC/mitron4-e.html or
http://www.assoc.tron.org/eng/document.html (It looks they need to
charge for the English version of the latest spec. to recuperate the
translation cost.)


As you can see, ITRON and T-Kernel is the specfication of API for a
realtime OS. It is open API architecture. ITRON has its open source
reference design from the very early days. There used to be an English
page at U. of Tokyo about the available ITRON implementatin that
conformst to somewhat older micro-ITRON V3 spec, but I can't find it
easily today. So I list the sample implementation offered at toppers
web site. (Toppers is NPO formed by users/developers of
ITRON-specification OSs.): http://www.toppers.jp/jsp-download.html
(Gee, I am afraid that the pages are available only in Japanese. The
above source can be used under BSD-like license, and/or GPL-style
license.)

T-Kernel has more strictly controlled source available at:
http://www.t-engine.org/T-Kernel/tkernel_e.html T-Kernel source code
is controlled by T-License. Read carefully before using it. OS based
on ITRON specification has been ported to many CPUs. See for exampe:
http://www.sakamura-lab.org/TRON/ITRON/list-e.html (I am afraid that
the list is outdated. I think there are ARM, PPC, and other CPUs being
supported.) T-Kernel has been ported to Renesas SH, MIPS architecture,
ARM architecture, and others.