A Deep Neural Network (DNN) Based Contract Policy on Hyperledger Fabric for Secure Internet of Things (IoTs)

IoT has revolutionized our interaction with the environment by enabling the connection of devices and sensors to the internet. However, the ubiquitous nature of IoT and the lack of a security framework result in various vulnerabilities particularly in data security. Blockchain of IoT (BIoT) has been studied recently to enhance security in IoT networks. However, this integration results in issues with network throughput, storage, and transaction security. To address these challenges, this paper proposes a deep neural network (DNN)-based contract policy to detect anomalous transactions upon their entry to the network and an encoding method to compress validated transactions. The proposed contract policy detects anomalies in IoT transactions, allows transactions to be stored in an encoded format, enhancing the security of an IoT network while minimizing storage space when storing IoT transactions. The proposed DNN-based algorithm rejected 2790 anomalous transactions out of 119,103 transactions and achieved a compression ratio of 47.81 for encoded transactions. Furthermore, we implemented the proposed solution on Hyperledger Fabric as an chaincode’s contract policy.