Anomaly Detection of DDoS Attack in Bitcoin Transactions using CNN-LSTM-Hybrid Model

Bitcoin, the initial use of blockchain technology, revolutionized online finance by providing a decentralized, open, and secure method of processing transactions. Blockchain's innovative architecture has influenced many industries by providing trustless data management systems and eliminating central authorities. While blockchain networks like Bitcoin have become ubiquitous and mainstream, they have also become more attractive targets for more sophisticated cyber attacks as well. Generalized attacks include 51% attacks, Sybil attacks, Distributed Denial of Service (DDoS) attacks, selfish mining, eclipse attacks, and time jacking. These attacks can seriously erode the stability of a network, data integrity, and user trust. DDoS attacks, in particular, aim at overloading the Bitcoin network with a staggering amount of malicious requests, thereby slowing down operations and genuine transactions. To enable identification and countermeasures for such threats effectively, deep learning techniques are under exploration. Convolutional Neural Networks (CNNs), Long Short-Term Memory (LSTM) networks, and combinations of CNN- LSTM models are promising solutions. CNNs are efficient at extracting spatial patterns, while LSTMs are efficient at handling sequential data. The CNN-LSTM model combines the two when they are applied in tandem to yield high accuracy real-time DDoS detection. Hybrid mode enhances responses and increases the ability to detect threats in complex blockchain systems like Bitcoin.