English

Abstract

This paper presents a novel method of user terminal positioning in cellular networks. A pattern matching technology based on received-signal-strength (RSS) from the pilot channels of cell towers has been the most popular network-based positioning method. In response to a position request, a terminal measures RSSs of pilot channels from surrounding cell towers and then the terminal’s RSS pattern is compared with a pattern database to find the most correlated one which indicates the position of the terminal. Although the pattern matching method can provide accurate positioning, its database construction and maintenance require a high overhead of periodic labor-intensive pattern collection. In this paper, we propose to exploit the call data records (CDRs) that are uploaded by Assisted Global Positioning System (AGPS) terminals as inputs to the pattern database, which removes or reduces the pattern collection overhead. In AGPS systems, terminals measure satellite signals and cellular network parameters (such as RSS) and relay them to the cellular infrastructure, which in turn calculates the terminal position using both that satellite and cellular network data. The proposed AGPS CDR based pattern matching method takes advantage of the increasing number of AGPS terminals in service: non-AGPS terminals can obtain more precise positioning results in areas where more AGPS calls are generated (e.g. hotspots). To do so, we analyze the characteristics of RSS patterns and AGPS CDRs. Based on the analysis, a pattern-distance metric and an AGPS CDR based pattern matching system are proposed and their performances are evaluated by examining field data of several urban downtown areas of Seoul, Korea. We obtain promising results: the position of the user terminal can be estimated with the accuracy (or, positioning error) at the level of 96.5m and 149.8m for the 67% and the 95% confidence interval, respectively.