Playing the "Standard" Game
Maybe we can sneak in some nice numbers, for fun," Trac Tran says of this story during a Monday afternoon interview over coffee in Xando's cafe.
This is the first time Tran, a coffee lover, has taken the time to walk through the nearby Baltimore Museum of Art's sculpture garden to the cafe on Charles Street. He keeps a cappuccino maker in his Homewood office. And the assistant professor has been pretty busy lately.
After Tran arrived at Hopkins in summer 1998, the young researcher starting exploring different mathematical "transforms" meant to speed up the transmission of digital multimedia--video, audio, etc. At the same time, he wanted to find a way to reduce the reliance on high-powered batteries or computer chips. His work could prove revolutionary: the electronic information industry is moving toward smaller and smaller devices. Yet these devices have limitations, especially power-wise, even as the need for more complex multimedia transmission is growing.
The future for this technology is, well, pretty futuristic: One should be able to view TV news or home security cameras, or get a video postcard from a daughter studying abroad--all on a PalmPilot, cellular phone, or as-yet-unheard-of Dick Tracy–type device.
Several months ago, the announcement of a promising transform Tran invented caught the eye of the technology and business press, as well as his engineering colleagues at various conferences, including the 1999 International Conference on Image Processing in Kobe, Japan. He has applied for a U.S. patent. Industry watchers call it a technology worth tracking. Cell phone users "could surf not only text-based web- sites, as they do today but also graphics-oriented ones that are currently too computationally intensive for handhelds," wrote R. Colin Johnson, contributing editor to the online technology publication, EETimes.com.
His process is actually a twist on current technology. To send video and images along the electronic highway, computers rely on compression--the squeezing of huge amounts of digital information into smaller packets. Trac's innovation involves a change in the mathematical process used to prepare the images before they're compressed and transmitted.
The math goes something like this: current image and video compression standards such as JPEG use what's known as the discrete cosine transform (DCT). (All DVD players and almost all PCs have built-in DCT modules either in hardware or software.) Yet DCT requires the costly, time-consuming multiplication of irrational numbers, the 1.768923-type numbers. Tran's method, which he has named BinDCT, translates the irrational numbers into "nicer" numbers: integers, 1, or 2, or 3--a sort of rounding off. These are then translated into binary numbers that the device can add more quickly.
Tran estimates that his method can transform and restore digital data three times faster than DCT. Since the complicated multiplication is absent, the demand for power is reduced, making BinDCT especially useful for handhelds, such as cellular phones and digital cameras. Those devices can't afford expensive computational power and rely on simpler binary addition.
The transform--raw technology that he completed in a research blitz over Christmas 1999--is one element of his broader research goal to create better technology for efficient data transmission to "make the world move faster." He says: "We don't want to do extra work if it's not needed. That's how smart human beings work."
Whether or not Tran's invention will ultimately win out in the "standard game" remains to be seen. Other image compression methods have a head start in the race to establish a standard for wireless multimedia transmission, though none has yet been established. The industry is focused on revised compression standards, such as JPEG-2000, which like others, still relies on the costly multiplication that's problematic in small devices. Yet the dilemma remains: better-funded, better-known technology could keep out Tran's innovations or others--just as VHS edged out BETA.
"The best technology doesn't always win. Other technology out there may be good enough and [more] easily incorporated," says Darren Lacey, technology licensing associate in Hopkins's Office of Technology Transfer, which is helping market and obtain patents for Tran's invention. "The standard game is really tough. There may be room for two to three technologies that do what his does, or maybe only one."
Applying for a patent also has its pros and cons. Companies might
balk at paying an inventor and university for a new process when
others exist. "If technology is patented you might be sabotaging
your ability to get incorporated into a standard," Lacey says.
"But I have a general feeling that is not likely to happen.
Patented technologies are now standards because people got behind
something and pushed it. If nothing else, you can articulate a
vision of how the technology is going to go."
RETURN TO JUNE 2000 TABLE OF CONTENTS.