Extended Indoor/Outdoor Location of Cellular Handsets Based on Received Signal Strength at Greenville, SC
Durgin, Gregory D.
MetadataShow full item record
This report documents an intensive, two-month measurement campaign exploring the performance of cellular handset location systems based on received signal strength (RSS). Building upon the success of the original Georgia Tech campus E911 location experiments, the new results in this report demonstrate the feasibility of RSS-based location in a 1920 MHz GSM network where the majority of calls originate indoors. Since most cell phone calls nowadays are believed to originate indoors (exactly where most proposed position location solutions fail), the ability to locate in-building E911 calls is a huge public safety problem. The results from this experiment are quite promising, however. The highest level of accuracy achieved by our location algorithm across the entire network was 51% of all test points having location error less than 100m and 79% having location error less than 300m. Although these numbers are slightly below the respective 67% and 95% safety targets set by the FCC, the Greenville trials represent a worst-case scenario for an RSS location system: a largely rural area with low density of base stations and a majority of indoor callers. The engineers from Georgia Tech’s Propagation Group compiled test measurements that included indoor data from 1 four-story hotel, 2 high-rise office buildings, 1 eight-story residential complex, 1 five-floor parking garage, 15 stand-alone restaurants, 23 small downtown shops, 1 grocery store, 2 department stores, and a variety of retail and shopping center structures. All of this data had to be painstakingly logged into georeferenced maps by hand since the GPS unit failed to acquire an indoor position 90% of the time. In addition to the extensive indoor measurements, the field engineers also collected numerous tracks of outdoor pedestrian test data and outdoor drive-test data. The end result is an extensive indoor/outdoor testbed for position location experiments that covers 63 square-kilometers of urban, suburban, and rural areas and contains nearly 90,000 measurement records. In arriving at the performance statistics, a number of position location innovations were made and documented along the way. These include: a new search-area limiting algorithm based on novel pieces of information in a network measurement report (NMR) (Section 6.1.3) an improved algorithm for estimating a handset location from a sequence of multiple NMRs (Section 6.1.4) new indoor penetration loss statistics for 1920 MHz (Section 4.1) demonstration that an RSS location system can still function after a major frequency plan change in the middle of a data collection (Section 5.1). Overall, the results of this experiment reveal interesting behavior of RSS-based position location that confirm the technology as an accurate, cost-effective way to improve public safety.