关键词: 室内定位, 射频指纹, 对传神经网络, 相关模型

Abstract: Pattern recognition algorithm based on location fingerprints database in RF has become one of the research hotspots in architecture-based network localization technology recently.A key issue of this technology is how to generate the database of “location fingerprinting”by using RSSI values received from the different access pointers.Without utilizing the spatial correlation feature of RSSI，the conventional method only samples the field directly，and results in large volumes of offline workload.In this paper a new method is proposed to generate the database by using Counter Propagation Networks（CPN） interpolation，which utilizes the spatial correlation feature of RSSI.The result of the experiment indicates that the method not only can achieve more accurate estimation，but also can greatly reduce the workload and save the training time.This method will make the fingerprinting technology more efficient and easier to implement.

Key words: indoor localization, RF fingerprinting, Counter Propagation Networks（CPN）, correlation model