An Enhanced Crow Search Inspired Feature Selection Technique for Intrusion Detection Based Wireless Network System

Recent development of cognitive computing driven evolutionary techniques improve the overall quality of service and user experience in wireless communication network. This Paper consists of a feature selection method based on improvement of Crow Search Algorithm which has been used in Intrusion Detection System to limit the size of the dataset with which the system is working with and getting better results. Since IDS deals with a large data, the crucial task of IDS is to keep efficient features which represents the whole data and there is no duplicity and irrelevancy. The previous model that was proposed used the crow search algorithm in the intrusion detection system (CSA-IDS) as a model to find the optimal feature’s subset and random forest as a judgement on features that are produced by the CSA-IDS. The KDD and UNSW datasets are used to evaluate the earlier proposed model. The proposed model achieved an accuracy of 99.84% for attack detection using UNSW datasets. Similarly, R2L and U2R attacks have detected accuracy of 99.97% for NSL-KDD dataset. The development of proposed model improve the overall communication services and feature selection in wireless communication network. The outcome proves that the subset of features that are obtained by using CSA-IDS fetches higher accuracy rate using a smaller number of features.