Joint activity recognition and indoor localization with WiFi sensing based on multi-view fusion strategy

Joint activity recognition and indoor localization is a promising technique, which is useful in smart homes or other IoT application scenarios. This joint task for the activity and location of a human could be estimated by channel state information (CSI) of Wi-Fi signal. However, there are now still shortcomings in the CSI processing schemes that could not well take advantage of the amplitude and phase information simultaneously. In this paper, a new scheme of joint activity recognition and indoor localization is proposed, which innovatively adopt a multi-view fusion strategy to perform wireless perception. Amplitude and phase information of multi-antenna are fully utilized to improve the recognition accuracy of joint tasks. Through the comparative analyses of experimental results, the proposed scheme has significant enhancement on recognition accuracy in joint tasks.