To detect the normal position of a mobile object by means of learning with pattern discrimination even under the abnormal transmission environment of a radio wave.

A service area is divided into plural zones (A-N). Base stations corresponding to the zones or a sensor station circulating among the zones are provided. The values of received electric field levels for the respective zones of the radio wave emitted from a mobile station are collected through the stations and they are made into one detection pattern. The detection patterns are sequentially obtained by changing the positions of the mobile station. Then, a database is generated while the correspondence between the detection patterns (drawings (a) and (b) and the positions of the mobile station at that time is learned. The database is retrieved with the detection pattern obtained at the time of detecting the position of the mobile station. When the matched detection patterns exist, the position of the mobile station, which is made to correspond on the database, is detected as the true position and is displayed (drawing (c)). When the matched detection pattern does not exist, the position of the mobile station is detected and displayed by the obtained detection pattern itself.