News Archive (1999-2012) | 2013-current at LinuxGizmos | Current Tech News Portal |    About   

Article: An interview with IBM’s Linux Wrist Watch project leader

Aug 8, 2000 — by Rick Lehrbaum — from the LinuxDevices Archive — 2 views

Earlier this week, IBM unveiled details about a fascinating research project taking place at IBM's research labs: a Linux-based wrist watch, which looks like this.

This morning, I spoke with Alex Morrow, “IBM Fellow” at IBM's TJ Watson Research Center in Yorktown Heights, New York, and project leader of the IBM Linux Wrist Watch project. My goal was to learn about the project's overall mission and goals, and also to gather technical specifics about the device itself, from a hardware, software, and systems design perspective.

RL: What is IBM trying to accomplish through this Linux Wrist Watch project?

Morrow: We think this is a breakthrough in Linux, getting Linux down into this size device. It's a good example of what we're trying to do, which is to see how broadly we can use Linux, and to demonstrate our commitment to having open platforms. Did you also see the news about [Linux] going up into the Blue Gene computer, with Linux?

RL: You're referring to IBM's recent statement about extending Linux “from wrist watches to petaflop super computers”? Yes, I saw that. But please tell me, what's a “petaflop”?

Morrow: It's one quadrillion operations per second. The system will consist of a million processors running simultaneously. It will be used to simulate protein folding at the molecular level. Basically, it's a huge [computing] device, with a million parallel processors, solving this particular problem in genetics.

RL: Do you actually have a million processors running right now?

Morrow: They will, by the time they're done. But that's absolutely what they're building. (info here)

Followup note: In checking the above website for info on Blue Gene, I could not locate an explicit reference to Linux. I therefore asked IBM to comment on the possible use of Linux in the million-processor application and received this statement from Ambuj Goyal, vice president of IBM Research: “We're considering using Linux for the Blue Gene supercomputer. We will experiment with that possibility.” –RL

RL: So, let's talk about IBM's strong fascination with Linux.

Morrow: You look at all this stuff, and you say “Why Linux?” The use of Linux makes it easier for us to work with both other software companies and individuals. Especially, with individuals in research institutions like colleges and universities. They can look at the device and say, “it's a cool device, I'd like to put some applications on it.” Things that we may never have thought of. And we can say, “ok, we'll use your knowledge of Linux and its tools to create the applications for the watch, and even to change the operating system for the watch.” So it's a great research tool, because it's so well known and, of course, the source is available for anyone to use.

RL: Is IBM Research in the mode of doing advanced development and then spinning it off into IBM funded startups? Is this a key part of the goal of IBM research, to spin technologies off to create products, divisions, or even whole companies?

Morrow: It's certainly not beyond the realm of possibility. However, in this case, the project really is about pushing the limits of technology and seeing what we can find out by solving the problems associated with building such a small device with constraints of form-factor, including the user interface issues . . . and the issues of communication. For example: how do you power the RF needed to communicate with other devices in the area? There is a whole bunch of interesting problems you wind up with when you build such a small device. It's generating a lot of IP (intellectual property) inside IBM. So, we're more interested in this as a prototype and as a learning vehicle, than as something we would actually manufacture and sell. However, that doesn't mean that if someone came forward with a business proposition, we wouldn't work with them.

RL: You're pushing the limits of Linux. The implication of this is that there's enough going on to take advantage of what Linux and related GNU software have to offer. For example: TCP/IP, GUI, and so on.

Morrow: We're actually running a full X11 (R6.3) on the watch, on the 96 x 112 pixel LCD display!

RL: The watch isn't a thin client, there isn't a server sitting somewhere hosting its applications, right?

Morrow: This is a real Linux system. We had a demonstration for the team in which we hooked up a TTY emulator through the serial port — which is part of the watch — and you could actually enter Linux commands and see them print back to you. It's a full Linux system.

RL: Are you running an IBM PowerPC CPU in there, or what are you using?

Morrow: It's actually not an IBM PowerPC, at this point, although that's certainly something we're thinking about for the future, for a low powered device. But it's an ARM7, right now.

RL: Is it an IBM developed part, or one of the off-the-shelf, commercially available system-on-chip devices?

Morrow: It's a standard Cirrus part (an EP7211), which is based on an ARM7 core plus additional built-in system controllers and interfaces.

RL: As for the Linux implementation that you're running, did your team do it all on their own, or did they start from a uClinux (or other) port for the Cirrus device, or what?

Morrow: They did start from the port for Cirrus, yes. It's based on release 2.2.1 of the ARM Linux kernel. We created a home-brewed boot loader.

RL: How much RAM and ROM are in the watch?

Morrow: 8M of each. The flash is read/write. It's a DiskOnChip.

RL: I understand the internal “motherboard” is just 27.5 x 35.3 mm in size. That's not a lot of space for all the functionality that's in the watch!

Morrow: There's not a lot of space in there — that's the phenomenal thing. The motherboard is quite nicely done. It's a very interesting board, a 6-layer board, all surface mount. It's roughly the size of a large postage stamp.

RL: You have a touch sensitive display. Given its small size, how do you handle the touch input, from a UI (user interface) perspective?

Morrow: We're probably only going to use four or five [touch input] points, for the user interface. We've also talked about having some plastic pushbuttons, but we haven't done that yet. Keeping it simple is very important to us.

RL: Over the past several decades, IBM has been an active player in both Unix and DOS. How does this relate to IBM's growing investment in — and emphasis on — Linux?

Morrow: DOS is actually an alley that we went into because of the IBM PC. Unix has been around for a long time, and IBM has been involved with Unix for quite a while. Unix and mainframes went together originally. Unix was on minicomputers and mainframes and it got on to the workstation because it was available. It's a great system to deal with RF, communications, TCP/IP, event processing, graphical user environments. The surprising thing is that you can now make [Linux] so small that you can actually use it in the form-factor of a wearable device. The thought of having the same operating system running on all the platforms that you want to deal with is very attractive. And doing it on the pervasive platform, or wearable platform, is even more attractive. It's a sort of level playing field for the ISVs and partners that want to have work with us. The nice thing about Linux is that there's a huge community of people who understand [it].

RL: It almost seems like Linux is, in a sense, a technology almost like C or C++, that can do practically anything you want. Isn't that what you're demonstrating here with the wrist watch project?

Morrow: I think it's almost as much a community as a technology. It's certainly a technology, but it's also the fact that people understand each other when they talk about the things they'd like to do to perturb Linux to do something different. You can pull a team together and say, “Wouldn't it be great if we could turn this roller wheel into a mouse?”, and they know what you mean by writing a driver to turn the roller wheel into a mouse. It's a community [consisting of] a large number of people who enjoy pushing the limits, and if you can find a way to do it with technology they understand, they get incredibly excited and they go off and solve problems that you never imagined could be solved. That's what's happened here.

RL: So, one of the benefits of “community” seems to be shared creativity, and shared technology components.

Morrow: Right. Exactly. There's the community of people who know Linux because it's an open source environment. They can get the sources, and they can talk to each other about specific things they need to do.

RL: How could any proprietary technology hope to compete with that, except in specific niche applications?

Morrow: It's an interesting question, and one that's very much at the center of this discussion of open source vs. proprietary source software. You're absolutely right. That's a good question. I think it's still an open question, but the fact is, the thing that's going to make a difference is the number of people who get excited about working on a technology. For us, right now, people are very excited about working on technologies where they can get access to all the source. And so here we have a situation where we can do that on a very tiny device, which is in itself exciting, so it's a doubly exciting project. When we show the watch to students, they get incredibly exited about it, and they want to go to work on it immediately. So this is a project which, by itself, seems a little strange — but it's really a project that pushes the limits of what you can do with computing.

RL: In terms of applications, what can you run, today, on the watch?

Morrow: Today, we're running clock technology. Various clocks. We also have a personal information manager which looks like a clock face and uses the roller wheel to let you look around to see what appointments you have, and lets you add new ones. Actually, that technology isn't running on the watch today. It's running on the watch development board now, and we'll probably have it running on the watch in about a month. We're also planning to put a couple of games on it. One interesting thing is that, at this particular screen size, you can see a face fairly well, so we are thinking about including faces in the information you store in your address book. We're experimenting, you might say, with different applications that you might want to run on the watch to make it something that would be more than a watch — almost like a little PDA on your wrist.

RL: And it could be an adjunct to your phone?

Morrow: Absolutely! We're looking at what it might be like to have this watch be part of a voice communications environment.

RL: Which brings Bluetooth to mind.

Morrow: Yes. Actually, there's a separate Bluetooth module that we've developed and that goes inside, along with the watch motherboard, inside the same case. There's a primary board with the CPU and associated stuff; there's a secondary board that has Bluetooth and the roller wheel. There are two versions of the secondary board, for the two versions of the wireless communications — Bluetooth, and a 900 MHz RF version. 802.11 takes too much power, around a Watt.

RL: How soon do you expect to build additional prototypes so that you can proliferate them around for developers to get their hands on them?

Morrow: We're deciding that right now. We're waiting to see the initial reactions to the announcements. It's something we're strongly considering.

RL: How do you enter data into the watch?

Morrow: Now you're getting into the issue of the user interface to the watch. The roller wheel (on the right side, next to the display — see photo), is basically used for scrolling forwards, backwards, and click-to-select (as on a mouse) from a menu of options. Or, you can use it to set the time on the watch, much as you would use the knob on the side of an ordinary (analog) watch. Also, it's a way you can enter text — although that takes a little while to do. Actually, I think a lot of text entry will be done through external keyboards. We are working on some other approaches, as well. One of the things that happens when you get to a form-factor that no one has ever used before, is that you start getting a lot of good ideas about how to solve problems like entering text on the device. For example, the touch screen could be used to enter text. We aren't sure yet whether that's what we are going to do (i.e. text entry via the touch screen), but that's one notion.

RL: Does the watch have any built-in sound capability?

Morrow: Yes. It has a microphone and a tiny speaker.

RL: Have you developed any communications capabilities yet for the IrDA port?

Morrow: Not yet. We're certainly going to do that. Right now, we're basically announcing the fact that we've actually got Linux running on this little guy. There will be more announcements as we get more things working on it. For example, we're looking at a higher resolution screen, at the communications capabilities. Taking advantage of Bluetooth and RF technologies, for example.

RL: Are there any other input/output ports or external connections?

Morrow: There's also a serial port that you can hook a couple of wires to, and use it as a TTY port. And there's a hole for reset, which will probably become a button.

RL: What sort of battery life does the watch have?

Morrow: Well, we were running it yesterday afternoon and by the time I left, which was four or five hours of operation at that point, it hadn't run out yet. We still have a fair amount of work to do in the area of power management. The battery is rechargeable, of course, although at this point you have to take it out to recharge it. We plan to have a stand that you put the watch on, to recharge its battery.

RL: How many researchers worked on the Linux Wrist Watch project?

Morrow: IBM Research has about 8 or 9 labs worldwide. We've been using a number of them. The fabrication was done at our Tokyo lab. The Linux work was done in India. The user interface design and overall project management were done here in Yorktown Heights. Zurich is looking at a higher resolution display for the device. Actually, one thing that's really exciting about this, is that we do have a worldwide team producing the watch. It's interesting to see how much a little device like this can really pull a worldwide team together. Overall, there have been roughly three or four dozen people working on the project, although not all full time.

RL: Thanks very much!



Related stories:
Fact sheet (& photos): IBM Linux wrist watch prototype
IBM develops prototype of wrist watch running Linux



Do you have questions or comments on this story? talkback here

 
This article was originally published on LinuxDevices.com and has been donated to the open source community by QuinStreet Inc. Please visit LinuxToday.com for up-to-date news and articles about Linux and open source.



Comments are closed.