<p>The rapid growth of IoT networks is projected to surpass 30 billion devices by 2030. This increases the threat of Distributed Denial-of-Service (DDoS) attacks targeting IoT devices with limited power, memory, and bandwidth. Although most previous research has explored deep learning and blockchain separately, few offer an integrated and lightweight solution to detect and prevent DDoS attacks in IoT networks. This paper introduces an integrated tamper-proof framework, HyDR—a hybrid deep learning and blockchain framework for real-time IoT DDoS resilience. HyDR combines three layers that incorporate hybrid deep learning (CNN-LSTM-Transformer), blockchain, and Interplanetary File System (IPFS) for secure detection of DDoS and decentralized logging. Trained in the ToN_IoT dataset, the IDS(Intrusion Detection System) achieved 99.01% accuracy with 0.990 precision, recall, and F1-score. The smart contract governed the interactions of the IoT devices, achieving a maximum throughput of 1.65 transactions per second. The latency measurements ranged from 256.88 ms to 2559.07 ms, reflecting the system’s responsiveness under varying operational loads. To ensure scalability and tamper-resistant data management, IPFS was integrated as the decentralized storage layer. Although throughput decreased during high-intensity operations, the framework consistently delivered real-time intrusion detection, robust access control, and scalable performance, demonstrating its viability for secure large-scale IoT deployments.</p>