Quantum Long Sort-Term Memory-based Identification of Distributed Denial of Service Attacks

Distributed Denial of Service (DDoS) is defined as an attack targeting a server, where the server is overwhelmed by an overflow of traffic, leading to a computer system being compromised. To prevent these attacks from happening, thus allowing the computer systems to maintain their normal behavior, a long short-term memory (LSTM) Time-Series model can recognize sequential characteristics in the data in order to predict whether a DDoS attack is occurring. Our approach leverages various aspects of network traffic such as Flow ID, Source IP, Destination IP, and Destination Port, and analyzes the accuracy and reliability using LSTMs to improve DDoS attack prediction accuracy. In this study we analyzed the relative efficiency of Quantum LSTM (QLSTM) over that of Classical LSTM, and found that both Quantum and Classical LSTM models train quickly and within a few epochs the accuracy increases to 0.99 and above. The validation of the models proved that QLSTM achieved better Recall, Accuracy, Precision, and F1 scores as compared to classical LSTM, making it an ideal candidate for DDoS attack predictions. This study proves the potential of applying deep learning techniques to strengthen network security, which has become increasingly important with the rise of complex cybersecurity threats.