An improved Kalman filter using ANN-based learning module to predict transaction throughput of blockchain network in clinical trials

【Author】 Hang, Lei; Ullah, Israr; Yang, Jun; Chen, Chun

【Source】PEER-TO-PEER NETWORKING AND APPLICATIONS

【影响因子】3.488

【Abstract】Clinical trials have been made transparent and accessible because of the widespread adoption of blockchain technology. Its distinctive characteristics, such as data immutability and transparency, could increase public trust in a fair and transparent manner among all stakeholders. However, blockchain systems cannot handle the requirement of processing huge volumes of data in real time. Scalability becomes a severe issue when implementing decentralized applications for clinical studies. With an abrupt expansion in the number of transaction exchanges happening consistently and the capital associated with those exchanges, there is an urgent demand for developers and users to know blockchain systems' performance limits to determine if requirements can be fulfilled; however, little is known about the prediction of blockchain system behaviors. This paper shows the feasibility of using machine learning technologies to predict the transaction throughput of blockchain-based systems in clinical trials. A learning to prediction model is proposed, in which the Kalman filter is used to predict the transaction throughput, and the Artificial Neural Network (ANN) is utilized to enhance the Kalman filter's prediction accuracy. A real dataset generated from a clinical trial testbed using Hyperledger Fabric is utilized to demonstrate the feasibility of the proposed approach. Moreover, we compare the Kalman filter with other learning modules, and the results indicate that the ANN performs best. Furthermore, we apply the proposed approach to different blockchain platforms, and the experiment results indicate the efficiency and universality of the designed approach.

【Keywords】Blockchain; Kalman filter; Artificial neural network; Learning to prediction; Transaction throughput; Clinical trials

【发表时间】

【收录时间】2023-01-08

【文献类型】实验仿真

【主题类别】

区块链技术-平台项目-Hyperledger Fabric