Wireless Technology Leaps the Last Mile
A US company has released a fast wireless technology that can be used to send data over the last mile—the final leg between communications service providers' facilities and their customers. GigaBeam's millimeter-wave technology could address the challenges found in accelerating last-mile communications.
This wireless approach could be important because building a wired infrastructure to customers, particularly for broadband services, is usually very expensive for service providers. For example, digging trenches for fiber is costly, said Doug Lockie, GigaBeam's cofounder and chief technology officer.
GigaBeam's millimeter-wave technology could enable a building on an optical wide area network to wirelessly provide high-speed communications to other buildings that don't have access to such services because they don't have their own optical fiber connections. All buildings would need roof-mounted millimeter-wave transceivers and antennas that are in direct line-of-sight with one another. In response, GigaBeam has developed proprietary circuits that provide radio functionality for transmitters, receivers, and other components that send wireless signals in the millimeter-wave band, which ranges from 30 to 300 GHz.
The US Federal Communications Commission licenses the transmissions—which have wavelengths that vary from 1 to 10 millimeters—in three frequency ranges within the larger band. The high bandwidth at these frequencies helps provide fast data transmission.
Rather than using silicon, GigaBeam's radio technology uses gallium arsenide or indium phosphide circuits, which have semiconductor properties that also enable higher data rates.
The data rates are comparable to those provided by optical fiber—1 Gbit per second—which is much faster than either T1 lines or cable-or DSL-based broadband.
GigaBeam's approach is a line-of-sight technology that requires an unimpeded path from sender to receiver. Communications originate from a radio transmitter and a two-foot dish antenna, according to Lockie. Recipients also need antennas, as well as radios with a fiber-optic connection for the data port.
"We need only one-millionth of a watt to connect a [1-Gbps] link," he said. Because the system can operate with so little electricity, he explained, it experiences few outages or other power-related problems.
Users can arrange GigaBeam systems in various configurations. For example, they might configure a system in a hub-and-spoke pattern so that a single Internet- or network-connected node could beam data to other points. Users could also create a full mesh network, in which all nodes could communicate directly with one another. This communications flexibility would make the system more robust and reliable, particularly if one or more nodes fail.
There have been other commercial, high-speed, wireless-connectivity technologies, but GigaBeam's approach offers higher data rates. Also, using licensed spectrum minimizes interference problems. And the narrow beams potentially enable systems to carry more transmission channels.
GigaBeam's current customers include corporations, municipalities, and universities.
Microwave Satellite Technologies uses GigaBeam's approach to offer telecommunications services to several New York City apartment buildings, hotels, and offices. Obtaining high speeds via a system that requires a one-time equipment purchase, rather than via service providers that charge ongoing fees, is a big advantage, said Microwave Satellite Technologies president Frank T. Matarazzo.
Millimeter-wave technology could appeal particularly to large organizations that want to connect multiple buildings on a campus affordably.
GigaBeam's approach could also be useful in the developing world, where there isn't a widespread wireline infrastructure, or in dense urban environments, where installing fiber or wires is cost-prohibitive, said Taha Rangwala, senior analyst with Pyramid Research, a market analysis firm.
However, Rangwala added, because the technology is new, many businesses that handle sensitive data, such as banks or hospitals, might be cautious about using it.
He predicted widespread deployment won't start for at least three years.
Researcher Turns Computers into Better Listeners
A University of Toronto associate professor has developed a way to make computers better listeners. Parham Aarabi has devised a method for extracting useful information from speech occurring in cluttered, noisy, or confusing environments, including those with multiple speakers.
"One of the major challenges with automatic speech recognition is the presence of noise," Aarabi explained. "Solving this problem is one of the key challenges in making speech recognition a reality for everyday and practical environments."
When speech-recognition systems detect a spoken word, they search through sound models they have developed and stored to identify it, he explained. However, correctly "hearing" words in the first place can be more difficult in noisy environments.
Aarabi created an algorithm that calculates the differences between the times at which various sounds reach closely spaced microphones and then estimates the locations from which the sounds came.
Users can then specify the general direction in which the system should "listen" so that it can focus on a particular source. They can also set the system to listen automatically to the loudest sound.
A key feature of the system is that the software partially filters out extraneous noises, which make them sound like a slight hum and greatly improves sound-recognition accuracy.
Once recognizing the appropriate speech, the system can search its database for the sounds that most closely match the ones it is "hearing."
The new technology will be most accurate when it is stationary in relation to speakers, such as would occur in a vehicle-based system. With cell phones, on the other hand, the user might move the handset in a random and unpredictable fashion, making the precise location of sound sources more difficult, according to Aarabi.
Multimicrophone systems will most likely be used first in cars with voice controls, he predicted. Other applications will include voice controls of consumer electronics products, speech recognition for devices such as tablet PCs, and speech input for PDAs.
However, Aarabi predicted, his system won't be ready for such complex applications for perhaps 10 years. "A great many challenges remain," said Aarabi. The first, he noted, is improving the filtering of extraneous sounds.
University of Toronto researchers have developed a system that uses multiple microphones and a specially designed algorithm to enable computers to focus on a particular speaker in a noisy environment, a typically troublesome task. In this figure, the system uses two microphones. The algorithm analyzes the timing of sounds reaching the microphones and determines the speakers' approximate locations. The system can then focus on either the speaker in the direction the user has indicated or the loudest speaker. Once the desired sound is captured, the system uses speech-recognition technology to figure out what was said. IBM Puts Encryption in a Processor
Scientists at IBM have developed a way to hardwire encryption technology into a microprocessor, promising more secure data storage as well as more effective digital-content protection.
The company says its SecureBlue technology would be useful in digital-media players, computers, PDAs, cell phones, and other systems that contain sensitive or copyrighted data.
SecureBlue adds functional cores consisting of just a few circuits each to a microprocessor. A representative implementation occupies only 2.5 percent of a typical microprocessor's area, noted Guerney Hunt, IBM Research's senior manager for distributed infrastructures.
Without using additional processor resources, the circuits can handle functions such as on-the-fly encryption and decryption, integrity checking (which detects whether an unauthorized user has changed or tampered with protected data), and transparent encryption-key generation and storage.
"Designers can pick and choose among the functions they want to include in their devices," Hunt noted.
The technology would sit between a chip's data cache and a host system's ROM, RAM, or flash memory.
SecureBlue encrypts the data selected by the user as it is written to memory. It decrypts the data only if a user enters the correct password upon logging onto a protected system. Thus, the data would be inaccessible if the system is hacked, stolen, or lost.
SecureBlue lets users leave some data—such as information that networked systems must be able to access automatically—unencrypted.
Owners of music, video, or other digital content could also utilize SecureBlue to limit the way consumers copy, distribute, or otherwise use the material. IBM didn't discuss details of how this digital-content-protection technology would work. However, many systems use encryption, along with other approaches, to limit content usage.
There are other hardware-based security systems, such as the Trusted Platform Module, which the Trusted Computing Group (www.trustedcomputinggroup.org) developed to store digital keys, digital certificates, and passwords. However, SecureBlue is one of the first hardware-based automatic encryption technologies.
The primary challenge of hard-wiring encryption technology into a microprocessor is making the circuitry fast enough to process material with minimal delay, noted Hunt.
"I think this technology will be very useful for several applications, particularly for securing sensitive data stored in mobile products," commented Dan Olds, principal with the Gabriel Consulting Group. "Hardly a day goes by without news of a lost or stolen laptop possibly compromising hundreds or thousands of [confidential or sensitive] records."
"I would expect early adopters of this technology to be government agencies, followed by finance, healthcare, and other corporate customers," he added. "But I don't believe this is a massive market in the short to [intermediate] term. This is a very high-end product for the high-end market."
News Briefs written by
Linda Dailey Paulson, a freelance technology writer based in Ventura, California. Contact her at ldpaulson@yahoo.com.
DOWNLOAD PDF 
