Indoor Localization with CSI Fingerprint Utilizing Depthwise Separable Convolution Neural Network

The WiFi-based localization approach has been widely used in the indoor environment. This paper proposes a MultIple Fingerprints-based Indoor localization system (MIFI). MIFI is based on the depthwise separable convolution neural network technique and utilizes Unmanned Aerial Vehicle (UAV) to help with transmitting fingerprint data. With the help of UAV, human effort can be decreased. In the training phase, we collect the Channel State Information (CSI) of the reference points. In the testing phase, CSI sent at the test locations are collected by Raspberry PI 4 as the input, then the system will output the predicted location. The experiment results show that MIFI can achieve a higher classification accuracy and mean localization distance error than the baseline work. Compared to the CSI data sent from UAV, only a minor performance is lost due to the drift problems of UAV.