A Hybrid Framework for Effective Intrusion Detection System in Wireless Networks

Communication technologies have evolved into new dimensions capable of providing data access to consumers at will and at high transfer rates. A novel hybrid scheme of IDS has been proposed and implemented in this research work to detect intrusions particularly DoS, U2R, R2L and PROBE attacks which are notoriously capable of collapsing the entire communication network. The hybrid combination consists of utilization of the well-known and highly efficient classification model namely the support vector machine (SVM) to detect the presence/absence of intrusions together with a nature inspired optimization algorithm, namely, the Glow Worm Optimization (GWO) capable of an optimal choice of features. The performance of SVM in the proposed work greatly depends on the fine choice of features extracted from the standard NSL-KDD dataset. Performance metrics for validating the proposed work include the classification accuracy, precision, recall, sensitivity, and specificity. These metrics have been compared against recent benchmark methods to justify the superiority of proposed work.