A HYBRID DQN–BLOCKCHAIN–PBFT FRAMEWORK FOR SECURE DATA TRANSMISSION AND REAL-TIME OPTIMIZATION IN SMART INDUSTRIAL NETWORKS
Dinesh Rajassekharan, P. Vidyasri, Vijaya Krishna Sonthi, Firas Tayseer Ayasrah, S. Arun Jayakar, M. Ramya, S. Maheswari, Sudhakar Sengan
Abstract: This research proposes a method that combines DQN + BCT + PBFT Model (Deep Q-Network + Blockchain Technology + Practical Byzantine Fault Tolerance) to optimise real-time smart industrial networks and enable secure data transmission. The model uses smart contracts and a distributed ledger (Blockchain Technology, BCT) for security. DQN supports adaptive decision-making and resource allocation. The model also includes a modified consensus mechanism for industrial applications. This model was compared with other methods, including Zero-Trust + LSTM, Hyperledger + Q-Learning, Ethereum + Neural Network, and Internet of Things (IoT). At a 256 KB block size, the proposed hybrid DQN + BCT + PBFT achieved a Transaction Success Rate (TSR) of 99.23% and a transaction processing speed of 87.43 TPS. Under a transaction load of 100 TPS, the proposed model achieved a Transaction-Load Response Time (TLRT) of 197.34 ms, demonstrating lower response time than several blockchain-based baseline models. Under a workload of 2000 transactions/min, CPU utilization reached 78.54% and memory consumption reached 4234.89 MB. NT analysis reported an average rate of 782.43 transactions per 5 minutes and a 13.4% performance loss during extended operations.
Keywords: Blockchain; Deep Q-learning; Security; resource utilization; Smart contracts; Internet of Things; Adaptive optimization
DOI: 10.24874/PES08.03.028
Recieved: 20.05.2026 Revised: 02.07.2026 Accepted: 31.07.2026
UDC:
Reads: 5 
