A Mamdani-Type Fuzzy Inference Framework for UAV Threat Detection and Countermeasure Deployment in FANETs

Abstract

The proliferation of Unmanned Aerial Vehicles (UAVs) in both civilian and military spheres has raised substantial security issues, especially within volatile communication environments such as Flying Ad Hoc Networks (FANETs). To challenge of the problem of real-time detection and response to UAV misbehaviour, the research proposed a Mamdani-type Fuzzy Inference System (MFIS) for real-time classification and detection of threats and subsequent actions. The MFIS is designed to take in information from UAV behaviours dataset (1000 samples) was obtained from Kaggle, consisting of four main features: energy consumption, mobility pattern, packet transmission, and link stability. After pre-processing the dataset through Min-Max normalization for standardization and amid Principal Component Analysis (PCA) for dimension reduction, the MFIS developed produces less computational load while retaining vital behavioural characteristics of the datasets. The results demonstrate the ability for the MFIS to enhance communication reliability while reducing key issues with routing and communication delays significantly over traditional FMIS methods like the Efficient Honesty-based Detection Scheme (EH-DS). The results show that the framework is an effective method for utilizing real-time context in making energy-efficient decisions for real-time UAV threat response. The simulation results show a significant improvement in performance parameters, including to-end delay, routing overhead (packets), and packet delivery ratio, by 15-55% compared to previous methodologies. While this framework has many advantages in terms of performance, these results confirm that the proposed fuzzy logic framework enables adaptive, accurate, and energy-efficient threat mitigation in real-time UAV operations.