A novel dilated weighted recurrent neural network (RNN)-based smart contract for secure sharing of big data in Ethereum blockchain using hybrid encryption schemes

Abstract

Background: With the enhanced data amount being created, it is significant to various organizations and their processing, and managing big data becomes a significant challenge for the managers of the data. The development of inexpensive and new computing systems and cloud computing sectors gave qualified industries to gather and retrieve the data very precisely however securely delivering data across the network with fewer overheads is a demanding work. In the decentralized framework, the big data sharing puts a burden on the internal nodes among the receiver and sender and also creates the congestion in network. The internal nodes that exist to redirect information may have inadequate buffer ability to momentarily take the information and again deliver it to the upcoming nodes that may create the occasional fault in the transmission of data and defeat frequently. Hence, the next node selection to deliver the data is tiresome work, thereby resulting in an enhancement in the total receiving period to allocate the information.

Methods: Blockchain is the primary distributed device with its own approach to trust. It constructs a reliable framework for decentralized control via multi-node data repetition. Blockchain is involved in offering a transparency to the application of transmission. A simultaneous multi-threading framework confirms quick data channeling to various network receivers in a very short time. Therefore, an advanced method to securely store and transfer the big data in a timely manner is developed in this work. A deep learning-based smart contract is initially designed. The dilated weighted recurrent neural network (DW-RNN) is used to design the smart contract for the Ethereum blockchain. With the aid of the DW-RNN model, the authentication of the user is verified before accessing the data in the Ethereum blockchain. If the authentication of the user is verified, then the smart contracts are assigned to the authorized user. The model uses elliptic Curve ElGamal cryptography (EC-EC), which is a combination of elliptic curve cryptography (ECC) and ElGamal encryption for better security, to make sure that big data transfers on the Ethereum blockchain are safe. The modified Al-Biruni earth radius search optimization (MBERSO) algorithm is used to make the best keys for this EC-EC encryption scheme. This algorithm manages keys efficiently and securely, which improves data security during blockchain operations.

Results: The processes of encryption facilitate the secure transmission of big data over the Ethereum blockchain. Experimental analysis is carried out to prove the efficacy and security offered by the suggested model in transferring big data over blockchain via smart contracts.

Keywords: Big data sharing; Blockchain; Dilated weighted Recurrent neural network; Elliptic Curve-Elgamal cryptography; Key optimization; Modified Al-Biruni earth radius search optimization; Smart contract.

Figure 1. The diagrammatic representation of blockchain security threats.

Figure 2. The suggested design.

Smart Contract icon image credit: Umeicon, https://www.flaticon.com/free-icon/smart-contract_7700640?term=smart+contract&page=1&position=2&origin=search&related_id=7700640, Flaticon license.

Figure 3. The MBERSO flowchart.

Figure 4. The structure of the recommended DW-RNN with MBERSO for the user authentication.

Figure 5. The smart contract based Ethereum blockchain using deep learning.

Figure 6. The recommended EC-EC approach with optimal key.

Figure 7. Convergence validation of MBERSO over various traditional optimization algorithms.

Figure 8. Precision analysis bar chart of MBERSO-DW-RNN in comparison to other methods.

Figure 9. Recall analysis bar chart of MBERSO-DW-RNN method with existing systems.

Figure 10. F1-score analysis bar chart of MBERSO-DW-RNN method with existing systems.

Figure 11. Accuracy analysis bar chart of MBERSO-DW-RNN method with existing systems.

Figure 12. MCC analysis bar chart of MBERSO-DW-RNN method with existing systems.

Figure 13. FPR analysis bar chart of MBERSO-DW-RNN method with existing systems.

Figure 14. Computational complexity analysis chart.