首页|Indoor Device-free Localization Using Received Signal Strength Indicator and Illuminance Sensor for Random-forest-based Fingerprint Technique
Indoor Device-free Localization Using Received Signal Strength Indicator and Illuminance Sensor for Random-forest-based Fingerprint Technique
扫码查看
点击上方二维码区域，可以放大扫码查看
原文链接
NETL
NSTL
Myu, Scientific Publishing Division
We presented an IDFL system utilizing RSSI and illuminance values for fingerprint localization and applied the random forest algorithm for pattern matching. From the localization training and testing results, random forest was found to have better precision, represented as STD, than other ML algorithms, i.e., k-NN and NN. However, k-NN was slightly better than random forest in terms of localization accuracy represented by RMSE. The localization results indicated that the overall performance is still relatively low. Unlike the device-based localization system, where the parameter values for fingerprint databases show larger differences, the effects of RSSI fluctuations on the IDFL system are similar between an empty and an occupied space. However, there is room for improvement by, for example, applying another signal parameter such as CSI, improving the data collection method, adding a number of datasets, and implementing deep learning in our future work. We expect that implementing deep learning and the use of CSI, which is more robust and reliable than RSSI, will improve the performance of IDFL.
NETL
NSTL
Myu, Scientific Publishing Division
