Indoor RADIO-MAP Construction Algorithm And Performance Analysis

Nowadays in the indoor wireless communication field the wireless local area network (WLAN) became a hotspot. The broad deployment and high speed of WLAN guarantees the WLAN based localization a broad prospect. In the widely used WLAN fingerprinting localization system, reliable, efficient RSS information is needed to make the system work. There are mainly two phases in the fingerprint algorithm: the off-line phase and the on-line phase. In the off-line phase, Radio-map is established and in the on-line phase, the localization system obtains the current RSS value and using the Radio-map to compute the estimated position.While most researchers in the world focus on the localization algorithm part of the system, this paper hold the view that as the basic of the system, the construction of Radio-map should not be ignored. To prove this point this paper presents a time-saving and accuracy-keeping Radio-map construction algorithm.First, the basic principles of the location fingerprint-based algorithm are introduced and the KNN algorithm is used to check the performance of different Radio-map. Then the stabilization of measured reference point's data is analyzed.In order to reduce the localization error by constructing a reliable Radio-map, an odd data filtering algorithm is proposed to remove the abnormal sample from RSS data. In this algorithm, all the RSS samples in the same reference point are considered as a whole and the distribution model of them is built to establish filtered Radio-map. The performance of original Radio-map and filtered Radio-map is verified by KNN. At the mean time, to fix the principal value of measured RSS samples, a grouping algorithm is proposed.Finally, to make the Radio-map establishing less time consuming, a grid interpolate Radio-map construction algorithm is proposed. Due to the loss of position information between the interpolated data and the measured data, the localization error increases. So to maintain the localization accuracy while using the interpolated Radio-map, an odd data filtering based grid interpolate Radio-map construction algorithm is proposed. Experimental results prove that this algorithm could completely satisfy application requirements and provide theoretical and technical supports to the establishment of WLAN indoor location systems.