NF-IDS: A Novel Framework for Intrusion Detection System
In current trends, networking has rapid development due to the tremendous utility of the security threats as major concerns. Thus the process of network security becomes a most essential thing. Many methods are used for ensuring network security. Among these Intrusion detection systems plays network services for communication purposes. This causes a significant role. Even though they had numerous benefits in secure the data. Still they lack in some issues like accurate classification of the intrusions, high computational complexity, false alarm rate are increased. To overcome these issues, a Novel Framework based Intrusion Detection System (NF-IDS) is proposed in this work. The motive of this work is to detect and classify the intrusion with enhanced accuracy. Here the contribution raw data is preprocessed to remove the unwanted noisy data. Formerly the structuresfrom preprocessed data are extracted using a novel Cosine Similarity based Principal Component Analysis (CSBPCA). Henceforth the optimal structures are chosen based on the Center Scalar based Modified Greedy-Backward Algorithm (CSM Greedy approach). Finally the types of the protocol traffics are categorized using a Modified Naïve Bayes Classification technique. The performance of the NF-IDS approach is validated using KDD Cup Dataset and compared with the several traditionalapproaches. The effectiveness of NF-IDS approach is proved with better classification accuracy with reduced computational cost and complexity.