Hafiz Muhammad Attaullah

Unmanned aerial vehicles (UAVs) in mission-critical applications are subjected to advanced cyberattacks of UAV-specific protocols and sensor streams, necessitating intrusion detection systems (IDS) that are sensitive to tight energy constraints and low-latency constraints on airborne edge systems. Current UAV IDS face a critical accuracy–efficiency trade-off: high-accuracy deep learning models exceed the computational capabilities of airborne edge platforms, while lightweight approaches fail to detect sophisticated attacks. This work introduces a novel Long Short-Term Memory (LSTM)-based framework that combines systematic multi-modal feature engineering with compact temporal modeling for resource-constrained UAV deployment. Our solution presents the Windowed Statistical–Spectral (WSS-64) Framework, a multi-modal feature intelligence system that ingests raw UAV telemetry and network traces and runs a 2 Hz compact LSTM model that converts these traces into 64-dimensional frames to provide consistent and accurate detection of classical cyberattacks, including spoofing, false data injection, and denial-of-service malfunctions. Our systematic combination of statistical, spectral, and network characteristics into fixed-dimensional representations optimized for a time model with resource constraints is the main novelty. The method yields a macro-F1 of 0.992 and an AUROC of 0.9995 with 99.21% accuracy and false positive (0.8%) and false negative (0.9%) rates, significantly higher than baseline methods such as Decision Tree (94.85%), Random Forest (96.32%), Support Vector Machine (95.10%), and Convolutional Neural Network (97.45%). The WSS-64 Framework offers O(N) linear complexity for real-time UAV operations. Received: 22 September 2025 | Revised: 25 November 2025 | Accepted: 6 January 2026 Conflicts of InterestThe authors declare that they have no conflicts of interest to this work. Data Availability Statement The data that support the findings of this study are openly available in IEEE DataPort at https://dx.doi.org/10.21227/6f22-py65, reference number [34]. Author Contribution Statement Hafiz Muhammad Attaullah: Conceptualization, Methodology, Software, Resources, Data curation, Writing – original draft, Writing – review & editing, Visualization. Inam Ullah Khan: Methodology, Resources, Data curation, Writing – review & editing, Visualization. Thabit Mahmood Thabit: Conceptualization, Formal analysis, Resources, Data curation, Writing – original draft, Writing – review & editing, Visualization. Muhammad Mansoor Alam: Validation, Formal analysis, Investigation, Resources, Data curation, Writing – review & editing, Supervision, Project administration. Mazliham Mohd Su'ud: Validation, Formal analysis, Investigation, Resources, Data curation, Writing – review & editing, Supervision, Project administration.

Unmanned Aerial Vehicles (UAVs) are emerging as key enablers of next-generation wireless networks, supporting flexible connectivity, cooperative sensing, and edge intelligence in envisioned 6G systems. However, the open, highly dynamic, and resource-constrained nature of UAV networks introduces significant security challenges that are poorly addressed by conventional centralized or static intrusion detection mechanisms. This article explores an Agentic AI-driven intrusion detection and self-healing security architecture; featuring autonomous agents with local perception-decision-action loops tailored for UAV-enabled wireless networks. By embedding autonomous intelligence, trust-aware collaboration, and adaptive decision-making across sensing, detection, and response layers, the proposed architecture enables localized anomaly detection, proportional mitigation, and autonomous recovery without continuous centralized control, achieving energy efficiency through localized inference and reduced communication overhead. A simulation-driven case study evaluates metrics like detection latency and accuracy, illustrating how agentic behavior evolves across interaction cycles, demonstrating adaptive trust degradation, bounded collaboration, timely mitigation, and effective self-healing while preserving mission continuity. Comparative performance characteristics further highlight the advantages of agentic security over static rule-based and centralized AI approaches in terms of responsiveness, adaptability, and energy efficiency. Rather than focusing on algorithm-specific optimization, this work emphasizes architectural design principles and system-level behavior, offering practical insights toward resilient, explainable, and scalable security solutions for UAV-enabled 6G networks.

As the deployment of Narrowband Internet of Things (NB-IoT) networks accelerates globally, their inherently constrained architecture and large-scale connectivity introduce critical security challenges. Hence, there is a need to present a penetration testing and risk assessment framework taxonomy specifically designed for NB-IoT ecosystems that would leverage ethical hacking methodologies. In this chapter, we conduct targeted vulnerability analysis using virtualized testbeds and industry-standard toolkits, simulating various attacks. We introduce a comprehensive risk taxonomy that categorizes threats across various NB-IoT network layers, including edge, core, device, and operational components. By mapping penetration testing results to this structured taxonomy, our study provides an effective means to evaluate risk severity, uncover security gaps, and recommend mitigation strategies. This work also addresses the ethical, and compliance aspects of penetration testing within IoT environments, ensuring responsible disclosure and alignment with recognized standards such as NIST and ETSI.

Unmanned Aerial Vehicles (UAVs) have become integral to modern consumer electronics (CE) ecosystem-based smart city infrastructures, providing dynamic connectivity, environmental monitoring, and situational awareness through distributed consumer Internet of Things (IoT) networks. However, their reliance on wireless communication links exposes them to a wide range of cyber and physical-layer attacks, demanding intelligent yet lightweight defense mechanisms. This paper presents a Multi-Agent Federated Edge Intrusion Detection System (UAV-IDS) designed to ensure privacy-preserving, real-time protection for UAV-enabled smart city IoT environments. The proposed framework integrates a hybrid Autoencoder–LSTM architecture deployed across edge nodes, where multiple intelligent agents collaboratively perform feature extraction, anomaly detection, and policy adaptation. A UAV-based coordinator manages federated learning rounds, aggregating model updates from distributed edge nodes without sharing raw data. Extensive experiments conducted on the T-ITS dataset demonstrate that the proposed system achieves 99.44% accuracy, 99.31% precision, and a 46.8 ms average detection latency, outperforming existing deep and federated IDS frameworks. Moreover, the lightweight deployment (8.42 MB) and low communication cost confirm the feasibility of the proposed design for embedded UAV operations. The results validate the framework’s potential as an adaptive, scalable, and privacy-conscious solution for next-generation smart city security infrastructures.

The growing deployment of uncrewed aerial vehicles (UAV) in autonomous and networked missions has heightened their exposure to both cyber and cyber-physical attacks, underscoring the need for intelligent and lightweight intrusion detection systems (IDS) solutions. This study introduces RoboLSTM-IDS, a deep anomaly-based framework that combines robust feature engineering with temporal sequence modeling for UAV network security. Leveraging Robust Optimization-Based Tabular Feature Engineering (ROBOTa), a robust optimization-based feature selection technique—the system extracts stable, high-impact features from complex UAV telemetry and communication data. These are modeled using a Long Short-Term Memory network to capture sequential attack dynamics. Comprehensive experiments conducted on five benchmark datasets, including real-world UAV cyber-physical data (T-ITS), CICIDS-2017, UNSW-NB15, and their CTGAN-augmented variants, demonstrate that RoboLSTM-IDS consistently outperforms traditional machine learning and deep learning baselines. It achieves up to 99.62% accuracy and 0.997 AUC, while maintaining low false positive rates and real-time execution performance. Unlike conventional IDS models that are computationally heavy, proposed model achieves a 6× smaller model size, 3× lower memory footprint, and significantly reduced inference latency. These results confirm RoboLSTM-IDS as an effective and scalable IDS solution tailored for next-generation UAV ecosystems.

Unmanned Aerial Vehicles are increasingly deployed in consumer applications such as logistics, disaster recovery, and surveillance, yet their wireless communication links remain highly vulnerable to cyber-attacks. Traditional intrusion detection systems (IDS) often struggle in UAV environments due to resource constraints, dynamic network conditions, and scarcity of labeled datasets. To address these challenges, we propose a Generative AI-enabled lightweight IDS framework tailored for UAV communication networks. The framework integrates hybrid Convolutional Neural Network–Gated Recurrent Unit (CNN-GRU) autoencoders with generative augmentation and knowledge distillation, achieving high accuracy while maintaining computational efficiency. Explainability is incorporated via SHapley Additive exPlanations (SHAP) analysis to ensure trustworthy and interpretable decision-making. Experimental evaluations on a multimodal UAV dataset demonstrate state-of-the-art performance with 99.49% accuracy, 99.48% recall, and AUROC of 0.9999, alongside a student model that reduces model size, inference latency, and memory footprint by nearly 50%. Comparative results confirm that the proposed framework outperforms recent IDS baselines in both detection capability and lightweight deployment, offering a practical and scalable solution for next-generation UAV communication networks.