Deep Belief Network Based Cyber Threat Detection with Blockchain Enabled Security in Private Cloud Environments

The risk of cyber threats in the context of private clouds is becoming more advanced and no longer conformable to rule-based and signature-based detection systems. In this paper, the authors suggest a hybrid model that integrates Deep Belief Networks (DBN) and Blockchain-Enabled Security Architecture (BESA) to detect automated cyber threats in real time in private cloud systems. The stacked Restricted Boltzmann Machines (RBMs) of DBN allow extracting hierarchical features in a deep manner out of network traffic to classify anomalies better than in handcrafted feature engineering. The blockchain tier offers threat recording, decentralized audit trails, and intolerance to tampering intelligence sharing amongst cloud nodes. Experiments on benchmark datasets such as UNSW-NB15, NSL-KDD and CIC-IDS2017 show that the detection accuracy is 98.7% with a precision of 97.9% and a recall of 98.3% and a higher F1-score of 98.1% than state-of-the-art baselines such as CNN-LSTM hybrids and federated intrusion detection systems. The proposed architecture dramatically lowers the false positive and the latency rates of the previous works and guarantees the auditability and data provenance of the multi-tenant deployment of a private cloud.