AI-Driven Behavioral Pattern Analysis for Enhancing IoT Security in Next-Generation Smart Cities

There has been a rapid development of 6G, SDN, and edge/cloud integration, which have simplified the implementation of IoT devices in the entire smart cities. These devices are frequently small, have a wide variety of different communication protocols, and can occur in large numbers, which contributes to their popularity as a subject of attack by cyber-attacks like DDoS, spoofing, and replay attacks. Traditional defenses such as encryption and firewalls are typically ineffective since they require excessive computer processing power and only respond once an attack has taken place. In this paper, the author proposes an active security system in which AI is used to monitor the behavior of devices and safeguard IoT systems in smart cities. We implement LSTM networks together with Isolation Forest algorithms to create a hybrid anomaly detector that trains on normal device behavior and alerts abnormality in real time. The system operates at the edge and latency and bandwidth are low. Experiments on real IoT traffic and simulated LoRaWAN networks demonstrate a 96.7 % detection rate with extremely low false alarms. The framework is policy-enforceable since it operates with SDN and network virtualization. The findings of our work prove that AI-based behavior modeling can be scaled, flexible, and suitable to protect next-generation networks and industrial IoT implementations.

Ҳали таржима қилинмаган