American Scientist

In 1993, an 18-year-old woman named Jennifer Koon was violently abducted at a shopping mall in Rochester, New York. She managed to call 911 from her hijacked car, but with no way to determine the geographic position of her cellular phone, police were unable to intervene during the 20 minutes that elapsed before she was shot to death by her attacker. Such tragedies helped spur the Federal Communications Commission to order wireless carriers to install equipment for locating the source of cellular phone calls to 911, a process that began in 1996. The last decade has seen considerable maturation of the relevant technology, including the ability of cell phones to use the satellite-based Global Positioning System. Still, the nation's cellular carriers are not likely to meet the FCC deadline for making such capabilities nearly universal by the end of 2005.

To understand why not demands a short review of regulatory history. The initial scheme for dealing with 911 calls from cell phones—Phase I in the FCC's plan—required wireless carriers only to provide the call-back number of the phone making the emergency call and the location of the cell tower with which it is communicating. Phase II mandated that cell-phone companies be able to locate a caller's position much more precisely, using one of two basic strategies. They can adopt what the FCC calls "a network solution," which demands a significant upgrade to the wireless infrastructure but typically allows subscribers to use standard phones. Or carriers can implement "a handset solution," which necessitates less modification to the network but requires that each user's phone contain special position-sensing hardware. According to the FCC, a network-based system must be capable of locating a caller to within 100 meters at least two-thirds of the time, whereas a handset-based fix must just as often be accurate to within 50 meters.

Many different approaches are possible. (A nice overview of these methods can be found, for example, in the March 7, 2002 issue of EDN magazine, found at http://www.edn.com/contents/images/198901.pdf.) The simplest, a network-based scheme, is called "time difference of arrival." It compares when the transmission from a phone reaches various towers. The difference in arrival times between two towers is enough to establish that the caller is positioned somewhere on a specific hyperbola. Two sets of differences (determined from three towers) fully locate the caller at the intersection of two hyperbolic curves.

Another network-based solution, called "angle of arrival," requires only two towers, but special antennas must be set up to determine the direction from which the caller's signals emanate. This information then allows the position of the source to be determined by simple triangulation.

Remarkably enough, a third network-based method exists that can, at least in principle, operate with just a single, standard cell tower. The technique is based on the fact that a phone's transmissions reach the receiving antenna not just once but many times. The first signal to arrive is the one that follows a direct path from phone to tower. Then come a series of signals that have reflected off buildings, power poles and the like. The peculiar mix of repetitions and delays received at the tower depends on the set of paths that the signals follow, which in turn depend on the location of the phone. Thus one can ascertain the caller's position by comparing the pattern of signals received with a library of "multipath fingerprints" and selecting the closest match.

Competing with such network-based approaches are the systems that rely on special-purpose hardware installed in each user's phone. One, known as "enhanced observed time difference," relies on synchronized signals transmitted by each tower. The phone uses those signals to determine its distance from various towers, information that it relays back to the network for calculating a position. Another handset-based solution is to outfit each phone so that it can sense signals from satellites of the Global Positioning System. The handsets differ from full-blown GPS receivers in that they alone cannot calculate positions, because some of the computational burden is shifted to the cellular network—hence the name "assisted GPS."

Whereas some carriers have adopted network-based solutions, Verizon Wireless and the newly merged Sprint Nextel Corporation (which together serve more than 40 percent of U.S. cell-phone users) use assisted GPS. Following FCC requirements, these companies must ensure that 95 percent of their phones are GPS-enabled by the end of this year. According to documents submitted to the commission, the management of Verizon concludes that "although it will come close to the 95% penetration level, it may not reach that milestone by December 31, 2005." Sprint's statements to the FCC are even more vague about what levels it expects to achieve. And Nextel is clearly running behind, with only half of its phones GPS-ready as of August 2005. As it explained in a recent report to the FCC, the company suffered a blow last year when it discovered a "software glitch which effectively dropped Nextel's A-GPS handset base to zero." It seems that a bug in the software used in all of its phones rendered them unable to process GPS signals received after July 18, 2004. Nextel has thus had to mount a massive campaign to get its customers to have their phones reprogrammed.

But even after all the wireless carriers get their networks and their customers' phones properly equipped, substantial difficulties will remain. For one, the FCC requirements pertain only to what positional information the phone companies must be able to provide—they don't mandate that the public-safety officials fielding calls to 911 have the equipment necessary to receive that information and to act on it. So even if location services on your phone and wireless network are all up and running, it'll be hit or miss as to whether the dispatcher on the other end of your emergency call can determine your position and send help to the right spot. (The same might be true even after businesses begin monitoring positional data from cell phones to hawk location-based services, a worrisome possibility that privacy advocates are now warning about.)

One reason is that the money for upgrades to their 911 systems is hard for some communities to come by. Although many states charge cellular-phone users a fee for improvements to  the wireless 911 system, these funds are sometimes used for other things. For example, New York diverts most of these monies to the state's general fund, for state police or, more recently, to various "homeland security" programs. Some New Yorkers are, however, working to straighten out this rather warped system. In particular, Assemblyman David R. Koon has battled against Governor George Pataki in an effort that get those funds spent on upgrades that would allow firefighters, police and paramedics to locate a cell phone being used to summon emergency aid. Koon is motivated by more than the usual ambitions of a politician: It was his daughter who died in 1993 when no one could locate the source of her wireless 911 call.