Transparent Threat Detection in Mobile Networks: A Real-Time IDS with Scapy

Abstract

: The increased reliance on wireless and mobile communication has intensified the need for practical, real-time cybersecurity measures. This paper presents a smart Intrusion Detection System (IDS) that inspects live network traffic and identifies malicious activity with minimal delay. Built on dynamic packet sniffing with Scapy, the system derives flow-level features and feeds them to a pre-trained deep learning model. Experiments demonstrate an overall detection accuracy exceeding 95%, average per-packet classification latency of 180 – 200 ms, dashboard update latency of 800 – 1000 ms, and LIME-based explanation generation of 150 – 250 ms per packet. Validation was conducted under benign conditions and simulated attacks (DoS attempts, nmap port scanning, spoofing, and ARP poisoning), maintaining a low false-positive rate throughout. A Flask-based web interface provides immediate visual feedback, while the use of LIME ensures interpretability by highlighting the most influential features behind each classification decision. The resulting system is adaptive, lightweight, and explainable for mobile and wireless environments, contributing to more secure and open next-generation wireless infrastructures.