Abstract: The openness of UAV Command and Control Link (C2) link makes it vulnerable to non-granted attacks, leading to the risk of UAV loss of control, crash and even malicious attacks on third parties. In this paper, in view of the risk of Denial of Service (DoS) attacks in C2 link and considering the lack of actual detection dataset conditions, we propose an attack detection method based on the Bidi-rectional Temporal Convolutional Network (BiTCN), which is a multi-source feature fusion capability, to construct a detection da-taset through timestamp alignment and forward data. Bidirectional Temporal Convolutional Network (BiTCN) attack detection meth-od, based on the fusion of information features of network data and physical data to construct the detection dataset, and solve the asynchronous problem of network and physical data through timestamp alignment and forward padding; the BiTCN model captures the information of the data in the forward and backward contexts through the bidirectional mechanism to capture the data's forward and backward information, complete the feature extraction and classification, and realize the detection of DoS attack, which is vali-dated on a real drone attack dataset.The experimental results show that, compared with the traditional FNN and 1D-CNN detection models, the proposed method has an accuracy rate (98.5%), precision rate (99.9%), recall rate (97.0%), F1 score (99.3%) and AUC (0.998) are better than the comparison model, achieving effective detection of DoS attacks on C2 links.

Key words: UAV C2 links, DoS attack detection, Bidirectional temporal convolutional networks, Fusion of cyber and physical data, Attack Detection Method