We check out what’s cooking in the research labs of the world’s largest chipmaker during this year’s Research@Intel Day.

WHAT ties the Atom processor, WiMAX and vPro technology together? Yes, they were all pioneered by Intel Corporation, but more importantly, they all began life as projects in Intel’s research labs.

Over the years, the ongoing research at Intel’s research labs has resulted in a number of interesting products, sometimes quite different from the company’s normal production of microprocessors.

For example, the research that went into what would eventually become WiMAX was started in 1999 by Intel Architecture Labs (IAL) to develop a mobile telecommunications network that was superior to current cellular technology.

Unlike other divisions within Intel, Intel Research is probably the most low-profile, with research that could take from five to seven years to actually blossom into actual products on the market.

“In a sense, Research@Intel day represents a sort of coming out of the research facility and how it impacts the future of the company,” said Justin Rattner, Intel’s chief technology officer and Senior Fellow.

The Atom processor itself has had an interesting history at Intel, according to Rattner — it was originally conceived in 1999 at the Microprocessor Research Lab (MRL) as an extremely low-power processor, codenamed “Snocone,” that would be still be compatible with the current Intel processor architecture.

However, though a one-watt processor was designed as a proof of concept, the project sat on the shelves for a while until it was resurrected in 2004 where work started on the design of what would eventually be the Atom processor — marking nearly 10 years from concept to an actual product.

Rattner, who gave the opening keynote for Research@Intel day, then went on to introduce the current research going on at Intel labs.

The current areas of research currently undertaken range from areas such as health and environment to visual computing, mobility and even purely exploratory research.

Here is a selection of some of the more interesting research projects of the more than 70 showcased at Research@Intel day.

LINK-UP: Gupta demonstrating Cliffside, the technology meant to simplify the direct connection between various WiFi consumer devices and a notebook PC.

Codename: Cliffside

Having said that most research at Intel labs will take up to seven years to actually make it into shipping products, it’s kind of ironic to start with Cliffside, as it’s a technology that we’ll be seeing in Intel Centrino 2 products coming out within the next twelve months.

Cliffside is Intel’s answer to the problem of connecting various consumer WiFi-enabled devices like (WiFi-enabled cameras and portable music players) to a PC or a notebook PC.

The main problem with current WiFi technology is that there isn’t an easy way to connect up, say, a notebook PC to a WiFi camera wirelessly without having to first having them all join a wireless network via a wireless access point.

Cliffside, which is to be found in the next-generation WiFi chipsets from Intel, codenamed Shirley Peak, attempts to solve that by offering a Personal Area Network (PAN) similar to a Bluetooth PAN where the notebook will see every WiFi-enabled device within range and allow them to directly connect by merely clicking on “allow” in the list of devices shown within range.

The main advantage of Cliffside is that it offers the same convenience of associating devices as in Bluetooth technology, but with the high-speed connection of WiFi.

The other WiFi devices require no other modifications to connect to the notebook — they will merely see the notebook as a wireless access point, and if allowed, will connect to it to synchronise or transfer files.

Ashish Gupta, Intel’s mobile platforms group product manager, demonstrated this technology, which is both a hardware and software solution, to automatically connect a Nikon Coolpix WiFi digicam to a WiFi-enabled notebook to transfer images.

Also shown was a Microsoft Zune portable music player connecting to the notebook, automatically launching the synchronisation application and then synchronising music with the player.

The Robotic Bartender

Undertaken by Intel Research Pittsburgh, this project’s goal is to create a general purpose robot that works in the home, doing a multitude of tasks perhaps to help the old or disabled.

While this seems like a fairly straightforward idea, it’s really quite a challenge when it comes to robotics. Although industrial robots are already in use in factories, they have very little of the uncertainty factor as they perform jobs which are repetitive in a very controlled environment.

According to Siddhartha Srinivasa, a research scientist at Intel Research Pitssburgh, the problem with having a robot in the home is how to deal with uncertainty — such as how it deals with changing terrain, or how to recognise and pick up objects of different shapes and sizes in various locations.

The research here deals purely with the software side of things, and the robotic arm showcased at Research Day was cobbled together out of mostly standard parts, like webcams, various sensors and of course, a robotic arm.

The demonstration of the Robotic Bartender, as Intel calls it, showed off the ability for the robotic arm, which has a webcam built-in to the area that would be the palm, to look at a table and recognise that there are mugs on it.

The arm then would plan its approach a mug to avoid the other mugs, pick it up and put it into a dish drainer.

PRESS TOUR: Mario J Paniccia, Intel Fellow and director of the Photonics Technology Lab, talking about the challenges of designing the next-generation fibreoptic technology to transfer the huge amounts of data coming out the multi-core microprocessors of the future.

The Robotic Bartender also has a Segway transporter that can be connected to it to allow the robot to move around.

Tera-scale and the future of computing

On a more familiar front, the researchers at Intel have also been working out the future of computing and how multicore processing would benefit both consumers and businesses.

In Intel-speak, tera-scale computing refers to terabytes of data that must be handled by platforms capable of teraflops of computing performance, an over-arching effort to shape the future of Intel’s processors and platforms.

The primary benefit of tera-scale computing are possibilities that it will bring — such as artificial intelligence for cars, physics simulations and even video indexing.

At Intel’s Computing Research Program lab, journalists were treated to a demonstration of video search — where a multicore computer goes through every frame of video, identifying and indexing a football game so that a user can then make a query such as number of goals or even search for a specific player and have the computer instantly take you to a point in the game with the goal or player.

EIGHTY CORES: A motherboard running a single 80-core processor being tested at Intel's Computing Research Lab in Santa Clara. The cores here are not meant as a full-on processor, but merely to test out input/output performance on a machine with many cores.

This kind of indexing requires vast amounts of computing power and at the research lab, Intel already has a computer running 80 processor cores.

Although the cores themselves are quite simplified, they were produced more to test the I/O (input/output) system and how it would work with so many cores working at once.

With so much data going in and out of the system, Intel researchers are also looking into how to move this volume of data in and out of the processor as well and this is where the Silicon Photonics lab comes in.

Silicon photonics is essentially a miniaturisation of current fibreoptic technology — fibreoptics are currently thought to be the best method to move large amounts of data from one place to another (at rates of up to 20Gbps), but the real challenge for the researchers at Intel is how to miniaturise this technology and produce it with current CMOS-compatible processes.

Before that can happen there are quite a few hurdles to get over — Intel is currently working on how to produce tiny laser emitters using the current CMOS-process used to create CPUs, as well as working out how to make the fibreoptic connections (which usually require careful professional alignment) on the motherboard easy enough for an average consumer to do reliably.

CONTEXT AWARE: A Sony handheld PC with Intel's context engine, demonstrating how the built-in camera and location sensing can be used together with a computer to become a virtual tour guide.

Context-aware computing

Another interesting demonstration shown was the research on context-aware computing.

Just what is context-aware computing? The concept is best explained by an example: Say a person is touring China, and he goes sightseeing at a local monument. His mobile device should be able to pinpoint his location using GPS, and then when he turns on the device, it should be able to use the built-in camera in the device to recognise the monument or building he’s pointing it at and then pull relevant information from the Internet to give him some history and a virtual tour.

SMART TECH: Lester Memmott, senior software architect in the Software Pathfinding and Innovation group in Intel SSG, showing the power of context-aware computing. The handheld computer here recognises the user is in a meeting and will do a Power Point presentation that will automatically connect to a wireless projector.

According to Kevin Kahn, Intel Senior Fellow and director of the Communications Technology Lab in Intel, PDAs and mobile phones today already have the separate components to make this work, but that a lack of tight integration between these devices was the problem.

Intel’s work is on a so-called “context engine” which will bring these sensing devices on the mobile device together to understand the environment the user is in and configure itself to suit it.

Another demonstration was of a Mobile Internet Device (MID) that would recognise if a person is in a meeting and when he or she sits down, will automatically connect the device to the wireless projector as soon as the PowerPoint application is launched.

This is achieved by obtaining information from the built-in accelerometer in the MID and an RFID sensor within the projector and the device. 

Secure Digital Wallet

On the security front, Research@Intel day also saw a demonstration of the Intel Secure Digital Wallet (SDW), which aims to help users protect their personal data when making online transactions.

The main problem with today’s security measures in online transactions is that most present systems are still vulnerable to a malicious user remotely mirroring a user’s PC screen or even logging their keypresses.

Intel’s SDW technology utilises Intel’s Virtualisation (VT) technology to create a so-called “hyper visor” to isolate the HTTPS secure transaction page (including the graphics) into a separate compartment that only allows authenticated and validated code access, thus thwarting any mirroring attempts.

According to Dr Prashwant Dewan, research scientist at Intel’s Corporate Technology Group, virtualisation also allows, say, a virtual image of a keyboard which is mouse clickable and can sit anywhere on-screen — this element of unpredictability on the screen can be used to stop keyboard hardware sniffing attempts.

The most important part of the SDW technology is that it can be implemented without modifying the operating system kernel nor requiring any infrastructure changes on the merchant side, thus reducing cost and encouraging adoption.