Increasing internet of communication in hypernet connects all the heterogeneous communication medium into single network through 5G and 6G. Through huge communications, the privacy data are publicly transferred to the peer medium without preserving sensitive data. Such privacy preservation in the health sector is emerging to protect the network data transfer from attackers. Traditional intrusion detection systems (IDS) fail to analyze the intrusion behavior features to degrade the true positive rate leading to poor detection accuracy which affects the precision rate, recall rate and f1 measure intended poor accuracy. To resolve this problem, to propose a smart intelligence based intrusion detection system using Optimal Elephant Hardening Feature Selection (OEHOA) and Deep Generative Adversarial Neural Network Classifier (DGANNC) to detect the Intrusion effectively. Initially the darknet communication dataset is taken to preprocess and verify the actual range of the margin. Then traffic fag, intensive delay, drop, tolerance feature margins are evaluated to analyze with Behavior Pattern Intrusion Rate (BPIR). To select the important features using OEHOA to reduce the non-relation feature dimension. Finally the Deep Generative Adversarial Neural Network is applied to identify the feature affected intrusion class. The novelty begins to reduce the feature dimension to predict with actual intrusion margin to improve the true positive rate with behavior analysis. The proposed system achieves high detection accuracy as well in precision, recall, f1 score and redundant false rate with lower time complexity.