Smart Airport Radar: Multimodal AI Classification of Aerial Threats with Communication Link Performance Evaluation

The proliferation of small unmanned aerial vehicles (UAVs) near airports poses increasing risks to airspace safety and infrastructure security. This paper presents Smart Airport Radar, a simulation-based framework for classifying aerial threats — including drones, decoys, and birds — using multimodal AI features. The system emulates dynamic swarming behaviors and extracts five key descriptors — mean speed, heading variability, jerk, thermal signature, and radar cross-section (RCS) — to train a multiclass Support Vector Machine (SVM) classifier. Comparative analysis with a traditional RCS-based rule method shows the SVM achieving a classification accuracy of 93.33%, far outperforming the baseline at 20.00%. Radar-style trajectory visualizations and class-specific precision, recall, and F1-scores confirm the model’s robustness and interpretability. Beyond sensing and classification, the framework incorporates a communication link performance evaluation, analyzing classification accuracy under varying Signal-to-Noise Ratio (SNR) levels. Results reveal that maintaining link quality above 15 dB SNR preserves near-optimal detection performance, bridging radar sensing with wireless communication reliability. With minimal computational overhead, high adaptability, and strong cross-domain relevance, the proposed system offers a robust, explainable, and deployable solution for real-time perimeter defense in modern airport security infrastructures.