In October 1996, the United States Federal Communications Commission mandated the implementation of position systems for wireless 911 emergency callers (E-911) [29]. This service is designed to provide emergency rescue and response teams with the location of a cell phone emergency call, comparable to the traditional ``911'' service for regular phones. In the final phase, wireless carriers are required to estimate the caller's position with an accuracy of 125 m (RMS) in 67 percent of cases. The details have subsequently been subject to debate, but this initial requirement gives an indication of the expected accuracy. The location systems developed for the E-911 requirement have been widely regarded as an enabling technology for location-based services; therefore, we will regard this level of accuracy as useful.
We assume that clients communicate position information to a location server with very high precision; in other words, the network client actually provides an accurate location to the location server. Position determination can be implemented either on the client itself (e.g., GPS) or by the wireless service provider, for example through triangulation of the wireless signal (hybrid approaches are also possible). To our knowledge, mobile phone operators in the United States found it challenging to meet the E-911 accuracy requirements through the latter approach. Thus, GPS information is likely far more accurate and privacy sensitive. Location-based service providers access location information through the location server. The full system comprises a location information source, a wireless network, location servers, and LBS servers. In a typical system, location information is determined by a location information source such as a GPS receiver in a vehicle. It is then periodically transmitted through a cellular or wireless network to the location server. When a vehicle sends a message or request to an LBS, the service accesses the vehicle's current location information from the location server, which acts as a proxy or middleware agent.
Finally, this paper focuses on services that do not require the user to log in and or present any kind of identifying information at the application layer. We believe that such LBSs will become available analogous to free services over the Internet. However, it would be interesting to extend this research to pseudonymous LBSs, which would allow tailoring services to individual users, for example.
Table 1 presents a summary of the resulting requirements.
Weather phenomena and corresponding road conditions typically cover larger areas. In addition, most warnings and speed limits must be given well ahead of the hazardous conditions. Thus, highly accurate position information is not necessary; about 100m road segments should be a suitable resolution for most cases. Conditions also do not change very abruptly, thus updates with a few minutes delay can be tolerated. In order to detect a change in driving conditions the external application needs quasi-continuous access to location information.
Since this application collects longer-term statistics, information delay is not important and time accuracy requirements are low. For example, it would be useful to distinguish night and daylight situations or rush hour from mid-day traffic but not to collect information with second-resolution. Precise location information is crucial, however, to pinpoint dangerous spots such as intersections or pedestrian crossings.
Response times of these services are important, thus, this application requires high time accuracy. The location, however, can be transmitted with medium accuracy; about 100m accuracy should be sufficient for obtaining point-of-interest information and area maps. Location is revealed only sporadically, when the driver issues requests. If such systems are used for navigation, the location can be revealed much more frequently.