NIDS-Using-CICIDS2017-KDDCup1999-Dataset

We have developed an IDS using neural network and machine learning algorithms based on two commonly used datasets: CIC-IDS2017 & KDD Cup 1999. We have trained artificial neural network (ANN) and machine learning models, including random forest, decision tree, and KNN. We evaluated and compared the metrics to find the best model.
We tested the trained models by initiating attacks using custom payloads and Kali Linux tools (such as hping3). These packets will be captured by packet sniffer (packetsniff.sh), a shell script developed using tcpdump. This shell script generates a PCAP file as an output. This PCAP file will be fed into CICFlowMeter, which extracts the features from the packets. The application's output is a CSV format file with six columns labeled for each flow (FlowID, Source IP, DestinationIP, SourcePort, DestinationPort, and Protocol) with more than 80 network traffic analysis features. This will be fed as input to the developed model, predicting whether the flow is benign or malicious. When malicious network flow is detected, the user will be notified.
Latest commit: Updated README and reorganized the repository folders