<p>The IoT has rapidly grown and has changed traditional network connectivity to smart environments that are tightly connected. However, most IoT devices do not have proper security features and hence are susceptible to different types of cyber-attacks. To address these issues, this study has presented a unique design that incorporates Deep Learning (DL), Software Defined Networking (SDN) and Blockchain technology to improve IoT cyber security. The main research question is aimed at identifying the possible approach to creating a cyber-security system for IoT-based smart environments to guarantee data confidentiality, secure financial transactions and proper threat identification. The proposed methodology is a combination of several sophisticated methods. The SDN control plane integrates the Squeeze-Excitation (SE) based Bi-Directional Long Short-Term Memory (SE based Bi-LSTM) model with the Honey Badger Algorithm (HBA) for traffic control. Blockchain is there to enhance the credibility and safety of the data and the transactions within the network. The Synthetic Minority Over-Sampling Technique (SMOTE) is employed to handle the class imbalance problem in the dataset to increase the model’s performance on the imbalanced data. The Bi-LSTM-HBA model is trained and validated on CICIDS 2018 dataset which is a realistic dataset for analyzing and mitigating cyber threats. To assess the efficiency of the proposed Bi-LSTM-HBA model in detecting high and low-frequency cyber threats, it is compared with other classifiers including GRU and BiLSTM. The findings show that the proposed Bi-LSTM-HBA model provides the best performance measures of 99.55% accuracy, 99.36% precision, 99.44% recall and a 99.42% F1-score. From these results, it can be said that the suggested model is very efficient in detecting and preventing cyber threats and surpasses other benchmark classifiers. Therefore, the Bi-LSTM-HBA model is a novel improvement in improving the security of IoT networks.</p>