An Enhanced Hybrid Approach for Real-Time Application-Layer DDoS Detection and Resilience in MANET Environments

Mobile Ad hoc Networks (MANETs) face growing security threats from application-layer distributed denial-of-service (DDoS) attacks, which create severe performance and reliability problems. This research develops a complete hybrid framework that combines Fuzzy Extrapolation System (FES) with Brownian Motion-enhanced Harris Hawks Optimization (BM-HHO) and an Extended Simplified Pulse-Coupled Neural Network (X-SPCNN) classifier. The BM-HHO algorithm identifies key features that reduce the input space to enhance model performance. The X-SPCNN system performs exact attack detection while the FES component assesses node trustworthiness and contextual factors to direct adaptive mitigation strategies. The proposed solution undergoes validation using a simulated MANET dataset, which includes malicious traffic patterns that simulate actual attack behaviour and real-world traffic of the Darknet. The framework demonstrates consistent detection accuracy above 99.2% with low false positives while operating effectively under typical MANET resource constraints. The combination of bio-inspired optimization techniques with neural classification and fuzzy reasoning enables a robust, scalable defense system that matches the dynamic decentralized characteristics of MANETs.