| Although the current positioning technology represented by GPS has been widely used, under the indoor environment, the location-based service has put forward higher and more specific requirements on the real time, accuracy and economy of positioning technology, and its scope of application and user experience have undergone new changes.Whether it is for ordinary users, or for commercial companies,indoor location information has become the key information of daily life trajectory and the production and consumption,and indoor positioning technology is confronted with many problems in adapting to the complex and changeable indoor environment.This paper, from the three key positioning links of physical measurement, data processing and location calculation, aiming at four main problems, including the low efficiency of artificial offline signal fingerprint extraction, the large amount of noise of signal fingerprint feature extraction, the complicated calculations of position matching area and the low accuracy of the performance of online positioning algorithm, respectively explored the corresponding solution.Firstly, a new model for dynamic acquisition and updating of signal fingerprint is studied. This model solves the problem of low efficiency of constructing signal fingerprint database, making the indoor signal fingerprint positioning technology truly become the indoor positioning technology based on the existing network, the existing infrastructure and the mobile terminal to be positioned.Second, the preprocessing method of signal fingerprint based on the priority of stabilization is analyzed. The signal fingerprint preprocessing method can effectively filter out a large number of volatile and random data in the original signal fingerprint, and extract the RSS fingerprint with high stability and strong position characteristics.Third, this paper discusses the partitioning method of dictionary tree positioning sub-regions. The time complexity of this method is only related to the query depth of the dictionary tree, and has nothing to do with the size of the offline RSS fingerprint database.Using this method, we can realize the fast and efficient localization of partitioning the sub-regions, which avoids the randomness of the partitioning method of artificial or K-mean clustering area, and reduces the computational complexity of the online position matching algorithm.Fourth, an indoor signal fingerprint localization algorithm based on the fuzzy neural network is proposed. This algorithm combines with the preprocessing method of signal fingerprint based on the priority of stabilization and the partitioning method of dictionary tree positioning sub-regions, and integrates fuzzy reasoning with neural network, which not only completes the preprocessing of signal fingerprint and the localization of sub-origins effectively, but also solves the problems such as the parameter adjustment of membership function and the generation of fuzzy rules, and improves the efficiency of localization algorithm. |
