Target Motion Direction Recognition Based on RSSI Time Series

With the widespread application of wireless sensor networks in intelligent transportation, urban security, and other fields, the efficient identification of target movement directions has become a pressing challenge. To address this issue, this paper proposes a dual-channel analysis method based on deep learning for accurate identification of target movement directions. First, sensor nodes are deployed to collect dual-channel Received Signal Strength Indication (RSSI) signals generated by target movement. The mean and variance are calculated using a sliding window, and time-series feature maps are constructed through Z-Score normalization. A residual network (ResNet50) is then employed to automatically learn features related to movement direction. Experiments demonstrate that the ResNet50 model achieves an identification accuracy of 80.43% in indoor scenarios, surpassing the traditional VGG16 model by 10.86%. The residual structure effectively captures waveform amplitude variations associated with movement direction. This method provides a novel technical solution for applications such as traffic flow monitoring and abnormal behavior detection.