Research On Indoor Fingerprint Location Algorithm Based On Grey Prediction Model

With the popularity of intelligent terminals, various applications based on location service were emerging, the requirements of these applications have become increasingly demanding in positioning. In the complex indoor environments, how to obtain the terminal’s location information quickly and accurately had become the key of research targeting. Because of its low cost, expand easily, the technology of indoor position based-on WIFI had become the mainstream of indoor position technology.Fingerprint location algorithm is easy to implement and rarely relay on environment, so it becomes the focus of indoor position technology research. However, fingerprint location algorithm also has the problems of large amount of calculation and low positioning accuracy.This paper analyzes the fingerprint location algorithm, points out the cause of positioning process may cause a large amount of calculation, and positioning accuracy is not high. So Based on the grey prediction model’s region partition and weighted KNN, an improved location fingerprint location algorithm GWKNN is proposed. The improved position fingerprint location algorithm is mainly carried out from the following two aspects.First, Online matching stage, the grey prediction model is introduced into the fingerprint location algorithm, proposes a method on grey prediction model of divide area. The purpose of this method is to narrow the scale of location, which can reduce the similarity calculation. At fingerprint location algorithm online matching stage requires a lot of similarity calculation, so it exits a lot of calculation, and it is need long time to locate. In this paper, the grey prediction is introduced into the fingerprint location algorithm, based on the before the current time of three position coordinates sequence, by using grey prediction model to predict the blind node’s current position. Based on the prediction position, the current speed and the time difference to divide area, which can narrow the scale of location, so as to reduce the amount of calculation, reduce the time of position;Second, Online matching stage, an improved online positioning method with weight KNN is proposed. In order to ensure the accuracy, according to analysis of the characteristics of energy loss of RSS distance on on-line position phase, based on analysis of the characteristics of noise filtering value, now, can calculate the similarity between the location node and the reference node, then use the filtered vector similarity for weight calculation standard to determine the value of the parameter list. At last, use the weighted KNN to calculate the blind node position.At last, this paper do the simulation experiment between the GWKNN-fingerprint location algorithm and traditional fingerprint location algorithm. Experiments results show that when the number of AP is more, the granularity of the location fingerprint database is most fine, the traditional fingerprint location algorithm and GWKNN-fingerprint location algorithm will reduce more. However, under the same conditions, the localization error of the GWKNN-fingerprint location algorithm is greatly improved than of that of the traditional algorithm. When the number of reference nodes add, compared with the traditional fingerprint location algorithm, the change of average positioning time of GWKNN-fingerprint location algorithm’s less. In all, GWKNN-fingerprint location algorithm’s average location error is less than traditional fingerprint location algorithm, and GWKNN-fingerprint location algorithm’s average positioning time is shorter when the fingerprint database is large.