Dynamic Autoencoder Framework for Advanced Anomaly Detection in Sdn Security

Resumen

Dynamic AutoEncoder-based Anomaly Detection (DAEAD) is an edge machine learning algorithm, with extensive application to real-time SDN network anomaly detection. DAEAD, which is dynamically based on autoencoders, self-adapts and learns based on adaptive network traffic patterns in order to construct efficient detection with very low false positives. The model exploits error in reconstruction and entropy analysis for an anomaly score computation and thereby identifies cyber attacks like DDoS attacks, probing, and zero-day attacks. In contrast to conventional approaches, the DAEAD dynamically adjusts itself to detection threshold based on current network conditions in real time and tries to provide high accuracy according to changing network conditions. DAEAD performs better than Graph Neural Networks (GNN), Recurrent Neural Networks (RNN), Generative Adversarial Networks (GAN), and Principal Component Analysis (PCA) when compared on all the performance metrics such as Mean Squared Error (MSE), Anomaly Score (AS), False Positive Rate (FPR), and True Positive Rate (TPR). SDN telemetry integration with the hybrid deep learning model can improve real-time threat inspection and system performance. Its capacity to learn ensures it the capacity to learn to detect cancerous network anomalies that are far too severe for best-of-the-best cyber defense. Its flexibility ensures functioning even with ginormous SDN networks without accumulation of anomalies over time. DAEAD over other traditional models by its performance deterioration through accumulation overtime, increases detection considerably by way of its ability to self-improve. Application of entropy-based analysis and decision-making focus on network topology and data heterogeneity improves decision-making. Intrusive attack security in breaking the security policy is implemented by the algorithm. Adaptive tuning of DAEAD in accordance with the network environment ensures appropriate security policy at the cost of resistance. Adaptive thresholding, deep learning, and application of SDN telemetry support scalability as well as adaptability in security architecture. The result confirms that DAEAD is the highest performing and most accurate anomaly detector. DAEAD is the ideal solution for emerging SDN infrastructures, ensuring real-time intelligent security.