Electric-Funeral

A Combination of Software Defined Network (SDN) And A Multi-Layer Perceptron (MLP) Neural Network That Results In The Mitigation of DDoS Attacks.

References

A dynamic MLP-based DDoS attack detection method using feature selection and feedback

Deep Learning-based Slow DDoS Attack Detection in SDN-based Networks

SDN-Based Intrusion Detection System for Early Detection and Mitigation of DDoS Attacks

A Flexible SDN-Based Architecture for Identifying and Mitigating Low-Rate DDoS Attacks Using Machine Learning

Requirements

• python3
• pip
• rust
• cargo

Installation

setup.sh

Generating data

First start the controller in generate data mode:

./network_controller.py --gen-data

Then start the network in normal interactions training mode (this uses mininet so it will probably require root privileges to run):

./create_network --normal

Once done, train for the attack state. Start the controller in generate attack data mode:

./network_controller.py --attack --gen-data

Then start the network in attack interactions training mode:

./create_network --all-attack

Training the MLP

Simply run the following:

./network_controller.py --train

Run DDoS Mitigation

Start the controller in detection mode:

./network_controller.py --detect

Then start the network in attack and CLI mode:

./create_network --attack --cli

The user should be able to ping the attack target with the following command:

u0 ping t0