EagleCyberNest

Generative Adversarial Networks (GANs) in IoT Data Security

Generative Adversarial Networks (GANs) are being explored as a way to improve cybersecurity by strengthening Autonomous Intrusion Detection Systems (IDS). These systems help protect Internet of Things (IoT) networks from cyber-attacks, but they struggle when training data is unbalanced, making it harder to detect less common threats.

Effective machine learning models rely on datasets that capture enough patterns and context. However, imbalanced datasets can weaken IDS performance, leading to missed cyber threats. This research focuses on using GANs to generate synthetic data, helping to balance cybersecurity datasets and improve detection rates for a wider range of attacks.

By comparing GAN-augmented datasets with traditional balancing techniques like SMOTE and Random Oversampling, this study aims to show how GANs can significantly boost IDS accuracy. With a stronger, more adaptive IDS, IoT networks can become more secure against evolving cyber threats.

Hierarchical Federated Generative Learning in Secured Smart Home System for Elderly

Smart home technologies are becoming essential for supporting the independence and well-being of elderly individuals. However, as more IoT devices and AI-driven systems are integrated into these homes, cybersecurity risks are increasing, leaving vulnerable communities exposed to potential cyber threats.

This research focuses on creating a secure smart home environment by addressing cybersecurity challenges in senior living spaces. One key approach involves using federated deep neural networks, which allow encrypted models to be transmitted instead of sensitive personal data, reducing the risk of cyberattacks.

Beyond technology, the project also recognizes the lack of cybersecurity awareness among many elderly individuals. To address this, it aims to develop a comprehensive cybersecurity education system and a real-time monitoring platform specifically designed to protect seniors and their smart home networks. By enhancing both security measures and awareness, this research ensures that smart homes remain a safe and reliable solution for aging populations.

Deep Reinforcement Learning Strategy in Edge Video Analytics and Networking

Managing real-time video data efficiently is essential in environments where low latency and smart resource allocation are critical. Edge computing helps process video data closer to where it is captured, reducing delays, but traditional methods often struggle with limited bandwidth, processing delays, and inefficient resource use.

This research explores how deep reinforcement learning (DRL) can improve edge video analytics by enabling adaptive decision-making for tasks like video transmission, resource management, and task offloading. By using DRL, systems can dynamically adjust network parameters and resource distribution, making video processing faster and more efficient.

These advancements are especially valuable in smart cities, autonomous systems, and disaster response management, where real-time video data is crucial for making informed decisions. By integrating DRL, this research aims to create a more scalable, responsive, and intelligent approach to edge computing and video analytics.

NetPrompt: LLM-driven Programmable Network Policy Management & Optimization

Software-Defined Networking (SDN) requires adaptive policy generation to ensure satisfactory Quality of Service (QoS) and Quality of Experience (QoE) expectations under dynamic network conditions. While generative AI can potentially automate the optimization of network configuration, there is a lack of methods for AI-driven policy automation and enforcement, particularly in translating high-level network intent into suitable service function chains using P4 switch configurations without misconfigurations. In this project, we present a novel framework viz., NetPrompt that uses Large Language Models (LLMs) for automated and intent-driven policy generation in SDN in the context of a video streaming application. By integrating prompt engineering and structured model refinement, NetPrompt adaptively selects the appropriate LLM configuration to generate suitable P4 scripts that align with user requirements, such as dynamic QoS adaptation.