OPTIMIZED DEEP LEARNING FOR CYBER INTRUSION DETECTION AND SECURED COMMUNICATION IN MANET

The popularity of the Mobile Ad-Hoc Network increases with low expense solutions to the real time applications. The dynamic nature, limited centralized system with low bandwidth often susceptible to the security threads and became the hot topic. To protect the MANET system it is essential to implement a secured system. The purpose of this work is to design an intelligent Cyber Security System (CSS) by collecting the threads data from the MANET system. Initially, the data is pre-processed using the Min-Max normalization technique, followed by the feature selection with the newly approached Mother Optimization algorithm (MOA). The feature selection is effectuated for speedy classification of intrusions and prevent the attacks and therein significantly enhances the security. To provide the security to the MANET system the work proposes a novel Adaptive Deep Belief Network (ADBN). This classifies the attacks and normal data and prevents the system from falling for the threat. Simulation is performed in NS2 and analyzed the effectiveness of the proposed work with existing systems. Our proposed work effectively enhances the security of the MANET system and surpasses all the other works.