An Intelligent Blockchain based Framework for secured Cryptocurrency Exchanges to Detect Fraudulent Transactions

Abstract

Over the course of the past several years, the swift growth and widespread use of cryptocurrencies have altered the face of the financial industry, presenting both opportunities and concerns. Alongside the expansion of cryptocurrency exchanges, there has been a rise in the number of fraudulent activities that take place within these exchanges, notably in networks such as Bitcoin. In this work, we report the findings of an exhaustive examination into the identification of fraudulent transactions within these exchanges, with a primary emphasis on Bitcoin. In order to accurately identify potentially fraudulent transactions, we plan to utilize a variety of Machine Learning (ML) techniques and ensemble techniques. One of these methods is the hard-voted ensemble model, which has been noted for its exceptional 99 percent success rate in previous implementations. In addition, we investigate the possibility of incorporating the hybrid SMOTE data balancing system and LSTM for the processing of time-series and edge-based transactions, which would further improve the efficiency of the model that we have presented. The findings of our research make a significant contribution to the development of strong and interpretable models, which are an essential component in the process of strengthening the integrity and security of the ecosystem of cryptocurrency.