Automatic Defense Against Distributed Denial of Service Using Anomaly Based Method in Machine Learning
Riyan Hakak, Manzoor Ahmad · 2021 Third International Conference on Intelligent Communication Technologies and Virtual Mobile Networks (ICICV) · 2021
Security of the network is biggest challenge in modern world. Big companies like Google, Microsoft, Amazon etc. need to have their network secured all the time as they host public information on their servers. DDoS attacks poses a serious threat to many organizations. One need to be alert always to identify such attacks as they come from various distributed sources and every time hackers try to attack differently, so filtering by hand to avoid DDoS attack is a difficult task. Goal of this paper is to train a machine to classify the packets into malicious packet or legitimate packet. This is achieved by training our machine with KDD CUP dataset. This dataset is divided into two halves containing 20% of original dataset and 10% of original dataset which is used for training ANN and testing ANN, respectively. The dataset is chosen uniformly from the original dataset set so that normal and malicious instances are approximately same in number in each dataset.