Proposed Mitigation Framework for the Internet of Insecure Things

Intrusion detection systems IDS are increasingly utilizing machine learning methods. IDSs important tools for ensuring the security of network data and resources. The Internet Things (IoT) is an expanding intelligent machines sensors. However, they vulnerable to attackers because ubiquitous extensive IoT networks. Datasets from intrusion (IDS) have been analyzed deep methods such as Bidirectional long-short term memory (BiLSTM). This research presents BiLSTM framework with Principal Component Analysis PCA (PCA-LSTM-IDS). PCA-LSTM-IDS comprised two layers: extracting layer which using PCA, anomaly classification. Through a comparison study, effectiveness proposed approach task detecting attacks evaluated NSL-KDD dataset, solves many inherent problems KDD'99 dataset. Experimental results indicate that suggested framework, average accuracy 99%, outperforms alternative approaches specially when explained variance 90%.