Improved Coati Optimization Algorithm for Feature Selection Enabled Intrusion Detection in Internet of Things

Internet of Things (IoT) network combines physical substances like sensors, networks and electricals along software for gathering and exchanges information. Because of security actions lack, the network objects are susceptible for severe at-tacks. For addressing this, effective security algorithm to deal with threat and detect attacks is required. In this article, Improved Coati Optimization Algorithm (ICOA) is developed to feature selection in intrusion detection. Dataset utilized in article is UNSW-NB15 dataset and cleaning of data and normalization are utilized as pre-processing techniques. The ICOA algorithm is used as feature selection technique which selects relevant features effectively. Next, the selected features are given to Random Forest (RF) classifier to detect and classify the at-tacks. The integration of RF algorithm and ICOA enabled feature selection, resulted in much effective training process and offered the method that determines robust performance in NIDS. Evaluation of developed algorithm is assessed to metrics using accuracy, precision, recall. F1-score and detection rate. Developed algorithm acquired accuracy 99.55%, precision 94.62%, recall 93.91%, ft-score 94.78% and detection rate 99.83% that is effective than other existing methods like Neighborhood Search Based Particle Swarm Optimization (NSBPSO) technique.