System Uses Wi-Fi to Provide Location Services
A US company has developed software that uses IEEE 802.11 (Wi-Fi) wireless technology to determine a user's location. Skyhook Wireless has released its Wi-Fi Positioning System for application developers and mobile-device manufacturers to use in their products. The company plans to make WPS available for individuals to download in the future.
Skyhook Wireless's Wi-Fi Positioning System uses Wi-Fi wireless technology to determine a user's location. When a Skyhook user is in a neighborhood, the client scans the area for access points. For each access point, WPS identifies the signal strength and the media-access-control address, whose physical location is identified in a database. The system then uses a proprietary algorithm to calculate the user's location. Source: Skyhook Wireless Applications and devices, such as cellular phones, could use WPS to offer driving directions, provide 911 emergency services for Internet telephony, or enable local search services that determine where users are so that they can provide the locations of nearby businesses and services.
The technology could even help locate stolen equipment. Skyhook's first customer was CyberAngel, a laptop-recovery system.
Skyhook hired people to drive the streets of 25 major US cities with laptops to find Wi-Fi signals from public and private access points including those for individuals and wireless LANs. The company then built a database of the access points' locations, identified via the media-access-control (MAC) address of their wireless routers, and their locations.
When a Skyhook user is in a neighborhood, the client scans the area every 0.1 seconds for access points, explained Skyhook CEO Ted Morgan. The system reads each access point's MAC address and compares it to the database, thereby obtaining multiple location and signal-strength readings around the user, Morgan said.
By using a proprietary algorithm that gives more weight to stronger signals, WPS calculates the user's location.
"We can determine your location within 20 to 40 meters," said Morgan. And, he noted, the more Wi-Fi access points in an area, the more accurately the system can locate a device. Also, he noted, WPS software can determine when a user moves an access point to another location and update the system's database.
The well-known Global Positioning System's accuracy depends on the quality of the receiver, environmental conditions, how much time the user spends in a given location, and other factors, explained Craig Mathias, principal of Farpoint Group, a mobile-technology consultancy. Although GPS can be very precise under some circumstances, he said, "I generally assume it is accurate to a couple hundred feet."
GPS consists of 24 satellites and numerous land-based receivers. The satellites transmit signals containing data on their location and how long it takes the signals to reach the receivers. A receiver can thus determine how far it is from a satellite. GPS triangulates signals from three satellites to calculate a receiver's longitude and latitude.
However, Morgan said, unlike WPS, GPS doesn't perform well either inside buildings or outside where structures block satellite signals. GPS is thus less accurate in urban areas.
WPS, on the other hand, is dependent on Wi-Fi deployment and thus will be effective primarily in urban areas but, unlike GPS, not in less-populated areas, noted Mathias.
Morgan said Skyhook plans to drive through cities updating the database's baseline data annually and add information on some European cities in the near future.
Researchers Build Reconfigurable Supercomputer
UK scientists are building an experimental supercomputer with field-programmable gate arrays (FPGAs), rather than general-purpose microprocessors, that can be reconfigured to tackle various types of tasks.
Researchers with Edinburgh University's Edinburgh Parallel Computing Centre say this could usher in a new era of small, powerful, flexible, energy-efficient supercomputers.
FPGAs have a block of logic gates that users can program to reconfigure the circuitry in various ways so that the chips can work on different types of problems.
"[An FPGA] can address only one specific task at a time but can be reconfigured to address any computational task," explained Rob Baxter, researcher and project manager for the supercomputer project.
The researchers have begun building a 64-FPGA computer that can operate at 1 teraflop. The fastest machine on the current Top500 supercomputer list (www.top500.org/lists/plists.php?Y=2005&M=06), IBM's BlueGene/L, runs at 136.8 Tflops.
Users can't program conventional microprocessors for single tasks. In addition, the chips have become so complex that they use considerable energy and generate a lot of heat, which often requires additional cooling technology.
FPGAs, on the other hand, can deliver more performance for specific tasks while running at slower speeds, noted Mark Parsons, commercial director of the FPGA High-Performance Computing Alliance (FHPCA), a consortium of technology companies, public agencies, and academic organizations including the Edinburgh Parallel Computing Centre.
The FPGA's hardware-based approach applies more floating-point units to a problem at one time than traditional processors and thus executes multiple instructions per clock tick, explained Baxter. This makes the FPGAs faster than traditional processors, which typically execute one instruction per cycle, he said.
This approach will also enable the new supercomputer to run cooler and be 100 times more energy efficient than a traditional supercomputer with the same performance, according to Parsons.
The Edinburgh scientists say the new machine will be smaller because it can use a single FPGA to do a job for which traditional computers need multiple conventional processors. The new machine will be the size of eight PCs, one-tenth the size of a traditional 1-Tflop research supercomputer, according to Baxter.
Professor Mark Snir, chair of the Computer Science Department at the University of Illinois, Urbana-Champaign, said FPGA supercomputers will work better for problems for which conventional processors aren't optimized and for which almost all of the computation can be done on the chip, such as specialized signal processing. This would minimize the need for external communication, which would keep the link to the chip from becoming a bottleneck.
The FPGAs' hardware compilers are less well developed than those based in software, which makes compiling an algorithm into an FPGA a task, according to Snir. And, he explained, users generally cannot program an FPGA via a standard programming language such as C or Java.
He also said that making multiple FPGA chips work together efficiently is a challenge because they're not designed to optimize interchip communications.
Thus, according to Parsons, a key to the Edinburgh project will be the FHPCA's construction of an FPGA Parallel Toolkit that will let multiple FPGAs in a system communicate with one another and run parallel numerical applications.
Other issues include whether the new supercomputer can handle a broad range of problems and whether it is relatively easy to program, Snir said.
According to Parsons, researchers will thus work on ways to help users port their existing code to the machines, thereby eliminating the need to write new code. "Without this support, [adoption] will be very limited," he said.
Hard Drive Saves Energy by Working while Resting
Samsung is developing a hybrid hard-drive/flash-memory system that can record data while idling, a technique that could significantly reduce energy use in limited-power, battery-operated laptops.
Samsung is developing an energy-efficient hybrid hard-drive/flash-memory system that can record data while idling. The system begins recording incoming data to a flash-memory chip. When the chip is full, the system transfers the information to the drive, which spins only until it has all the data. Hard drives typically spin all the time, unless set to stop after a period of inactivity, noted Ivan Greenberg, Samsung Semiconductor's director of strategic marketing.
Thus, hard-drive motors are one of a laptop's biggest energy-consuming parts. With Samsung's new technology, the 2.5-inch drive rarely spins because it includes a 1-Gbit Samsung OneNAND flash-memory chip. It also works with a flash transition layer (FTL) that makes flash memory look like a disk drive to the operating system.
The system begins recording incoming data to the chip. The FTL identifies how much data the flash memory contains. When the flash device is full, the FTL communicates with the hard drive's firmware, then dumps the nonvolatile write buffer's contents onto the mechanical disk and erases the buffer. When the disk has all the data, it stops spinning.
The new drive thus typically spins only about 30 to 45 seconds every half hour. Samsung hopes to improve this to 30 to 45 seconds every hour. This could extend the life of a four-hour laptop battery by about 36 minutes, or about 15 percent.
The proposed hard-drive system looks interesting for consumer electronics but its usefulness would depend on how much cost it adds to the products, which are very price-sensitive, explained Jim Porter, owner of Disk/ Trend, a disk-drive-market-research firm. He said Samsung hasn't discussed the new hard drive's potential price. Porter estimated the new drive could add $100 to the cost of a laptop.
Greenberg said Samsung hasn't built a prototype system yet but has produced an emulation system and plans to construct engineering and commercial samples in the next year. He predicted laptops with the new hybrid drives will hit the market late next year.
News Briefs written by
Linda Dailey Paulson, a freelance technology writer based in Ventura, California. Contact her at ldpaulson@yahoo.com.
DOWNLOAD PDF 
