AI-Driven Deep Belief Network with Block chain-Quantum Framework for Intrusion Detection in Private Cloud

Internet of Things (IoT) devices grow by leaps and bounds and private cloud infrastructure becomes even more explosive, these two factors have created new sources of cyber risk that require intelligent, automated, and scalable detection systems. Traditional Intrusion Detection Systems (IDSs) have serious limitations in resisting zero-day attacks, polymorphic malware, and distributed Advanced Persistent Threats (APTs) in dynamic cloud-edge environments. This paper introduces a unified AI-Driven Deep Belief Network with Blockchain-Quantum (ADBN-BQ) framework, which combines five complementary ML and DL architectures: CNN-LSTM hybrid, Self-Attention BiLSTM, Random Forest, SVM and XGBoost, along with Deep Belief Networks (DBN) that include Restricted Boltzmann Machine (RBM) layers, Blockchain-immutable audit logs, and quantum-inspired Shabal feature permutation hashing. The pre-processing pipeline integrated SMOTE class balancing, PCA, and MI Feature Selection, reducing dimensionality by 40% and training time by 35%. Experiments on 5 benchmark datasets (NSL-KDD, BoT-IoT, TON_IoT, CICIDS-2018, UNSW-NB15) show that the detection rate is 99.2%, the F1 score is 98.9%, the false positive rate is 0.8%, and the block throughput is 4,200 transactions per second with a 380ms finality time, outperforming 12 state-of-the-art methods. The following physical validation results confirm the viability of real-world edge deployment: 28 ms inference time, 45 MB of memory, and 96.8% accuracy on a Raspberry Pi 4.