Summary: | 碩士 === 龍華科技大學 === 電機工程研究所 === 99 === The satellite in weak signal environments is not able to provide adequate functions, such as urban areas and indoors. Wireless signal location fingerprinting method meets the law of Spatial variability and Temporal consistency, it’s benefits from the complex signal propagation by using the unique characteristics of the wireless signal of certain calibration points to achieve the effect of indoor location estimation. In this work different deterministic approach to exploit the fingerprints in the location estimation phase are studied, the location method includes two phases : off-line phase and on-line phase. The off-line phase create a radio map and each calibration’s database by received the wireless signal strength values. The on-line phase uses the time step of one second to receive the instant wireless signal strength values and to do with radio map information to locate than the estimate by using the estimates algorithms : K-nearest neighbor algorithm and weight K-nearest neighbor algorithm to estimate the location effect of test, the general used is Euclidean distance and Mahalanobis distance to estimate the position’s distance, this paper presents effective combination of gray relational analysis and Mahalanobis distance with the distance estimation method and apply it in the weighted K-nearest neighbor algorithm to locate the weight in the estimate. The experimental test platform using the mobile phone of android system for verify the location results of this thesis, this feature can be extended to the android platform to develop location-based applications. Finally, through experimental results show that proposed in this paper combined with Mahalanobis distance and grey relational analysis location estimation method, can indeed achieve more accurate positioning results.
|