News

MICROELECTRONICS TECHNOLOGY ALERT

SEND MULTIMEDIA FASTER WITH LESS POWER

It's a mathematical breakthrough but not one that will dwell in the rarefied ivory towers where such high math moments usually linger. It will find very real applications in such things as laptops, digital cameras, and handheld computers. It will also find applications in less portable electronics image handlers such as video conferencing.

An electrical engineer at Johns Hopkins has a new digital transformation method that will speed the transmission of digital multimedia while using far less power. The speed alone is enough to make it easier to conduct high-quality video conferencing in real time. The lower power demand will make it more feasible to process images on battery-operated devices.

The mathematical breakthrough, BinDCT, lets a computer transform and restore digital data three times--or more-- faster than today's technology. In sending images and video from one location to another, computers use compression to squeeze large amounts of digital information into smaller packets that can be moved quickly electronically. Before compression, the data undergoes a mathematical transformation. The transformation is what Hopkins's Trac Tran has reinvented.

He changed the mathematical process used to prepare images before they are compressed and transmitted. Current transformations require multiplication and multiplication takes a lot of computational power and a big chip. BinDCT requires no multiplication at all. It uses, instead, binary shift and addition operations. These can be completed with a much smaller and less power hungry chip.

The transform is fast and binary friendly. The microchip can be very small with low power consumption. It maintains the same level of image quality as the larger chips using multiplication now do. It should significantly improve the data processing capability in compact electronic devices that rely on small chips and limited power supplies. That means you are talking about low-cost digital cameras, palm-size computers, and perhaps even cellular phones that can keep up with the multimedia communication rate. Trac Tran has applied for a US patent on the new BinDCT transform.

Details: Trac D. Tran, Assistant Professor, Dept. of Electrical and Computer Engineering, Johns Hopkins University, 207 Barton Hall, 3400 N. Charles St., Baltimore, MD 21218-2686. Phone: 410-516-7416. Fax: 410-16-5566. E-mail: ttran@ece.jhu.edu.