| Summary: | 碩士 === 元智大學 === 通訊工程學系 === 106 === This paper will discuss how to use the Raspberry Pi to create a low-power Bluetooth device (iBeacon) to obtain signal strength (RSSI). The system under consideration includes four low-power Bluetooth devices built by the Raspberry Pi and an esti- mote iBeacon and in the lab Eighty-four coordinate points were generated without obstacles. Data were collected one by one at each coordinate point for the low-power Bluetooth devices built by four Raspberry Pi’s and one estimote iBeacon. However, machine learning method-based neural network and the famous Gaussian Mixture model(GMM) were used to perform prediction on the data. Comparing the models in terms of error rate, MLP produced the lowest with 3%.
This paper first introduces the Raspberry Pi and iBeacon and explains the motiva- tion of using the Raspberry Pi as a Bluetooth sender and receiver and introduces the machine learning methods and neural networks used, as well as the experimental flow and results.
Keywords: Raspberry Pi, iBeacon, GMM, MLP
|
|---|
