The rapid expansion of the Internet of Things (IoT) has revolutionized urban living, allowing cities to become more efficient and interconnected. Yet, this connectivity also brings significant cybersecurity threats, particularly as malicious actors exploit vulnerabilities within these systems. To address these concerns, researchers have developed the Advanced Artificial Intelligence with Federated Learning Framework for Privacy-Preserving Cyberthreat Detection (AAIFLF-PPCD). This innovative model aims to bolster cyber resilience without compromising user data privacy.
IoT technology has become integral to modern smart cities, improving urban services such as transportation, waste management, and energy consumption. While the benefits of these technologies are substantial, they also create ripe opportunities for cybercriminals to exploit system weaknesses. Recent advancements, paired with the growing sophistication of cyberattacks, have prompted the need for advanced security measures.
The AAIFLF-PPCD model, developed by researchers at King Abdulaziz University (KAU) led by Mahmoud Ragab and his team, integrates advanced AI techniques with federated learning principles. The core method centers on collaborative data processing among IoT devices, allowing systems to learn from shared insights without requiring direct access to sensitive data. This characteristic is pivotal for maintaining the privacy of users within these smart ecosystems.
The AAIFLF-PPCD approach incorporates three integral stages: feature selection, attack recognition, and hyperparameter tuning. Utilizing the Harris Hawk optimization (HHO) method, the model first identifies and selects relevant features from the IoT data, ensuring computational efficiency by discarding irrelevant information. Following this, the stacked sparse auto-encoder (SSAE) classifier plays its role by detecting cyberthreats effectively through deep learning strategies.
Finally, the walrus optimization algorithm (WOA) is employed for hyperparameter tuning, fine-tuning the model for optimal performance. Collectively, these technologies yield significant results; the AAIFLF-PPCD technique boasts impressive detection accuracy rates of 99.47%, surpassing previous models.
Ragab emphasizes, "The AAIFLF-PPCD approach aims to preserve privacy without sacrificing the detection performance of cyberthreats.” This focus on privacy is particularly relevant, as public concern over data sharing within smart city frameworks continues to grow.
Alongside the impressive accuracy of the AAIFLF-PPCD model, its efficiency is also noteworthy. It achieves substantial performance improvements compared to other existing cybersecurity techniques within IoT settings, hence offering promise not only for current needs but also for future advancements as urban environments adapt and evolve.
The findings underline the potential of the AAIFLF-PPCD framework to provide scalable and effective cyber protection solutions for increasingly complex smart city infrastructures. Looking forward, continued research will be necessary to tackle the dynamic nature of cyber threats, ensuring secure and sustainable urban living.