Abstract
Traditional cloud storage solutions often suffer from deficiencies in fault tolerance, scalability, privacy, and security. Decentralized storage can address these concerns. However, the untrusted and potentially unreliable nature of distributed storage nodes in IoT and edge cloud settings requires careful selection of nodes for file storage in open and heterogeneous distributed solutions. We present here a secure and reliable distributed storage system with trust-based workflow scheduling, which takes into account a Dynamic Key Management and Distributed Encryption Network (DKMEN) for multilayered data security and integrity persistence, a novel Hybrid Encoding and Replication Network (HBERN) for improving file availability, and a multi-factor trust management system when making file placement decisions. The proposed system also employs blockchain technology for on-chain metadata storage for immutability and reliable accessibility. We evaluated the proposed system in real-world scenarios for diverse chunk loss environments on the InterPlanetary File System (IPFS) network. The results show that the proposed system offers a significant reduction in the probability of data loss compared to the state-of-the-art system Snarl. Moreover, the system guarantees data integrity and security by employing the DKMEN, a chunk-level multi-layered encryption algorithm, with each chunk encrypted independently applying a unique encryption key, protecting the data against vulnerabilities. Moreover, the trust management system addresses capacity and reliability as two central factors where capacity refers to the available resources and reliability considers the past node performance that determines the trustworthiness of the storage node. We experimentally demonstrated that this trust model enhances file placement decisions in distributed storage networks. Thus, the proposed system presents an effective solution to alleviate the key issues of data security and availability, providing better estimations of a node’s trustworthiness in distributed storage systems.