Abstract: DDoS assaults, which interrupt the availability of services across the board, are one of the most dangerous forms of cyberattacks now in existence. The complexity of DDoS detection stems from the fact that it must analyse a large amount of real-time traffic as well as a wide variety of attack methods. In this work, we introduce LUCID, a lightweight deep learning distributed denial-of-service (DDoS) detection system that uses Convolutional Neural Networks' (CNNs) inherent features to distinguish between malicious and benign traffic flows. Specifically, we add four things to the literature: (1) a novel application of a convolutional neural network (CNN) to detect DDoS traffic with low processing overhead; (2) a dataset-agnostic pre-processing mechanism to produce traffic observations for online attack detection; (3) an activation analysis to explain LUCID's DDoS classification; and (4) an empirical validation of the solution on a resource-constrained hardware platform. When tested on the most recent data available, LUCID's detection accuracy is on par with that of the state-of-the-art methods, but its processing time is cut in half. Through our evaluations, we show that the suggested method can effectively identify DDoS attacks even in contexts where resources are few.
Keywords: Distributed Denial of Service, Deep Learning, Convolutional Neural Networks, Edge Computing.
|
DOI:
10.17148/IJARCCE.2024.13632
