Comparative Analysis on Network Attack Prediction Used Deep Learning Approaches on Software Security Testing

Intrusion Detection Systems (IDS) are pivotal in identifying and mitigating security threats in IoT networks. Yet, they are frequently targeted by malicious attacks. This study compares the performance of Whale Integrated Long Short-Term Memory (WILS-NET-based IDS) and Dense Random Neural Network (DnRaNN) models in identifying these attacks. Both models are evaluated based on their capabilities and performance. WILS-NET-based IDS utilizes OMNET-python API for real-time IoT network simulation and behavior analysis of nodes. Established datasets like UNSWNB15, KDD, and CIDDS-001for comparative purposes. The DnRaNN approach employs ToN_IoT dataset and is assessed across various parameters. The comparative analysis indicates DnRaNN’s superiority, achieving 99.15% and 99.05% attack detection accuracy in binary and multiclass scenarios respectively, compared to WILS-NET’s 99%. This confirms DnRaNN’s effectiveness in accurately identifying diverse attacks in IoT networks.