BSIDM: secure IoT data management using blockchain, AW-DAG consensus, and HABE encryption

Today, more than 55 billion IoT devices generate about 79.4 zettabytes of data per year, highlighting the growing gap between centralized cloud architectures and the security, scalability, and access-control needs of Internet of Things (IoT) applications. At the moment, blockchain solutions for managing IoT data remain fragmented: lightweight blockchains neglect access control, permissionless DAGs fail to provide fine-grained access control, and permissioned blockchains trade off throughput for Byzantine fault tolerance. In this paper, we propose a BSIDM (Blockchain-enabled Secure IoT Data Management) architecture that consists of four planes: IoT Sensing, Edge Preprocessing, Blockchain Coordination, and Cloud Storage that communicate via smart contracts. BSIDM architecture utilizes two novel ideas: an AW-DAG reputation-weighted, DAG-based Byzantine Fault Tolerance consensus protocol with sub-second finality due to parallel tip validation, and Hierarchical Attribute-Based Encryption (HABE), a hierarchical ciphertext-policy attribute-based encryption with fine-grained access control in O(m + log k) complexity, where m is the number of attributes matched by the access policy and k is the number of authorities in the hierarchy, so decryption cost grows linearly with m and only adds a small O(log k) hierarchical-delegation term that can be applied to edge computing devices. Dynamic sharding through consistent hashing ensures cloud storage load balancing, and blockchain-secured migrations ensure auditability. The security of the BSIDM is formally proved, as it provides IND-CCA2 security, Byzantine fault tolerance for f < n/3, and prevention of Sybil attacks. The experiments performed on the 500-node CloudLab testbed, for comparison with IOTA Tangle, Hyperledger Fabric v2.4, the IPFS storage system, and AWS IoT Core, demonstrate that BSIDM achieves 5,847 TPS throughput, an average latency of 320 ms, and 0.8 s finality. Thus, BSIDM has a 5.97 × (5,847 vs. 980 TPS) improvement in throughput and a 90.1% reduction in average latency compared with IOTA, using just 0.23 mJ per transaction, 1.7% more than AWS IoT Core energy usage. Comparative analysis across nine aspects demonstrates BSIDM’s dominance.