A Secure Framework Based on Blockchain Methodologies to Identify Authentications in Decentralized Document Verification Systems

Abstract

Document verification is a process of evaluating the documents of end users or participants of the network and verifying the authentication in the most trustable way. The verification process was done manually in the olden days. Because of the development of technology, each document is converted into digital documents and faces issues like duplication, loss of records, data theft, and forgery of documents. These digital documents have been verified initially with their own identity. Identity authentication is very much essential to confirm the uniqueness of every individual. Many document verification systems exist globally, utilizing more security layers for verification methods. But they lacked in a few factors like privacy, transparency, high cost, and latency. Each verification method is controlled by a centralized authority. Schemes towards economically improving security with transparency and reduced processing time are required in decentralized systems adopted for storing and sharing documents. The relevant study focused here refers to pursuing a blockchain mechanism presented via three possible modules of implementation. This research focuses on improving security with transparency and reducing processing costs and time in a decentralized system. A blockchain mechanism is used for its decentralized nature that would be applied in storing and sharing records. The research is classified into three modules. The first module is concerned with verifying the possibility of anomaly detection using deep-learning (DL) algorithms, considering the academic record as an exemplar of input details. In this, academic records are taken as the input data. Relevant malicious activities (such as forging and/or document alterations) are detected via the Tensor Flow approach, wherein the processing time is saved in its hash values. Further, regarding any image details considered, the deeplearning algorithms with backpropagation suites are adopted to verify possible anomalies. Observed changes in hash values are indicators of