Skip to content

Mettings Summary

Jump to bottom
Zekun Li edited this page Oct 11, 2019 · 2 revisions

September 11:

Project: Detection of the distributed denial-of-service attack using a machine learning approach
Search for proposed work in

  • Google Scholar
  • ACM
  • IEEE

within 2015-2019
Keywords: "DDoS" AND "Machine Learning/Data mining/etc." Search for: Features, approach, result

September 20:

Accuracies sometimes are not true, focus on precision and recall.
Create a GitHub repository contains all the code and findings, create a wiki page.
Prepare reports of proposed work for their:

  • Features
  • Datasets(public) with link
  • Techniques
  • Type of DDoS they were working

The Project will focus on specific types of DDoS

October 4

Focus on making one model works well for one specific type of DDoS. Look for info about UDP lag.
General types of DDoS has been working for years, and cause Cloudflare and load balancing technique typical types of attack are not as useful.
Use UNB dataset.
CIC meter works on UNIX. It read network flow and generate additional features, but not all of them will be used in this project.
Process pipeline:

Preprocessing

  • Get pcap and use CIC meter to generate net flow
  • Remove unused features in the CSV and keep the raw features
  • Optimize the type of numerical value to reduce the size of the data
  • Create a Python function for feature generations

Get more normal traffic data

Clone this wiki locally