Detecting a distributed denial of service attack using a pre-processed convolutional neural network

This paper presents a scheme for detecting distributed denial of service (DDoS) attacks for smart grids. The main procedure of the proposed approach consists of applying a discrete wavelet transform to input data to extract features; training a convolutional neural network (CNN) to the extracted features; and testing the CNN to detect anomalous behavior in the data based on a threshold determined in the training parameters. The implementation detected the DDoS attack with 56.1% accuracy with the one stage CNN and 80.77% accuracy with the one stage pre-processed CNN.