Blockchain-Based Outsourced Privacy-Preserving Support Vector Machine Classification on IoT Data

With the enormous amount of IoT data generated, stored and exchanged, users are currently resorting to service providers to perform data mining classification tasks. Empirically, the IoT data used for training a classifier is moved to cloud servers running by third parties. Data integrity and the protection against data breaches is, therefore, becoming crucial. In this paper, we introduce BChainSVM: a blockchainized privacy-preserving support vector machine (SVM) classification between mutually distrustful IoT data owners and the cloud provider. With BChainSVM, we aim to bridge the gap between the theory of privacy-preserving classification in data mining and its practice. We first present the system model, design goals and construction details. Our security analysis indicates that BChainSVM achieves accurate and private classification, data correctness and data integrity. Furthermore, our performance evaluation shows that BChainSVM is feasible and efficient in terms of communication and computation costs.