The Research On Ap Selection And Location Determination In Fingerprint Location Method

The context-aware computing becomes more popular in recent years. The context is used to describe the condition of the entity. The entity is an enormous definition, and is includes various objects. The location is one of the most important condition, and the location-aware computing is a branch of the context-aware computing, so how to get the accurate location is becoming an important research subject. In outdoor location area, the Global Position System (GPS) and Cellular network can meet the demand of human, but both of them can not get a desirable result when they are implemented in indoor location area. In this paper, we mainly discuss the indoor location in wireless local area networks (WLAN), the research object is the selection of access point (AP) and the location determination method in Fingerprint location method.The location technology implemented in indoor location environment includes Infrared, Bluetooth, RFID, ultrasonic and WLAN, as the high speed channel, the facility of the implementation and the enormous cover of WLAN technology, that we adopt WLAN technology to realize the location. In the following part, we make a discuss of the various location method. The location method mainly includes range-based and the range-free location method, the ranged-based method mainly includes the geometry location methods, such as the trilateral and triangle method, compared to the ranged-based location method, the ranged-free location method do not requires the accurate distance measurement, and has a low power consumption and low cost, but a low accurate. It mainly includes centriod, DV-hop, APIT and Fingerprint method. Based on the comparison of the various methods, the Fingerprint method is most adapt to the WLAN location environment, and in this paper we make a research of the Fingerprint method.The main research contents of this paper are the AP selection method and the location determination method in Fingerprint method. The main function of AP selection method is to reduce the online location computation, and filter those APs whose function is not very good, so that raise the location accuracy. Currently, the AP selection methods are almost only to reduce the computation, and can not works to filter the APs. In this paper, we develop a robust online AP selection strategy for the indoor location tracking. It takes the environments changes into account and makes use of residuals ranking algorithm to select those APs least sensitive to the environment changes in indoor location tracking, such it not only can reduce the location computation, but also can raise the location accuracy, we call it ResidualRanking method. In the research on the location determination aspect, this paper makes a improvement on the customer Bayesian location method. The customer Bayesian location is the unique location classification based on the maximum posterior probability, so it can only make accuracy to the calibration level. The improved method makes the final location determination by weighting all the training samples, and this result is continuous.Based on the above research, we present a location tracking system called BRR (Bayesian and Residuals Ranking) which is based on the Bayesian decision method and the ResidualRanking method we proposed. This system also provides the basis platform for the related experimental analysis. In the same experimental condition, compared to the MaxMean method, the ResidualRanking method has about a 0.34m accuracy raise, the imporved the Bayesian method has an about 0.45 m accuracy raise compared to the classical Bayesian method. BRR location system has an about 1.92m average error, and the smallest error is 1.71m.