Summary: | 碩士 === 元智大學 === 通訊工程學系 === 104 === In recent years, the popularity of mobile devices, also led to the evolution of the built-in Bluetooth technology. In 2013, Apple announced that using smartphone iBeacon technology, that makes the use of the Wi-Fi to indoor localization technology is no longer ahead of us. iBeacon technology makes smartphone or others mobile devices execute commands within the Bluetooth sensing range. With Bluetooth devices, the application installed in a smartphone can probably find it, and the relative position of the Bluetooth device.
In this paper, we investigate how to use apps on smartphones with iBeacon to get signal strength. The system includes four Bluetooth devices and a smartphone. We design 154 measurement points in this environment, with 512 measured data pairs for each point and each Bluetooth device, resulting a total of 315, 392 data pairs. Moreover, we use different Machine Learning methods, including kNN(k Nearest Neighbors algorithm), SVM(Support Vector Machines)and LDA(Linear Discriminant Analysis), to predict the coordinates of points. This study uses the root mean square error(RMSE) to compare these three Machine Learning methods. Therefore, using the Machine Learning approach to perform indoor localization with high accuracy is an important research issue with great prospects.
keywords: iBeacon、Bluetooth low energy、RSSI、Machine Learning、RMSE
|