Packet-Filtering-in-SDNs-with-ML

Given a filter, only traffic deemed legitimate by this filter will be allowed.

Avoid incoming traffic from websites hosting unwanted content with a domain name which cannot does not indicate the same

Motivation

The current firewalls can easily be fooled, for example using content spoofing. Rules for dropping packets from website with a good domain name but hosting unwanted data, is rather difficult to set in traditional firewalls.

We aim to solve this problem using SDNs and ML.

Workflow

Setup

• Operating System = Ubuntu 18.04
• Programming Language = Python3
• APIs = beautifulsoup, bert embeddings, ryu, scapy
• Platform = Mininet / OpenFlow13.

Instructions

The variables which need to be modified for change in environment:

• firewall_monitor.py:24:: sys.path.insert(1, '/home/harsha') # for importing checkSim.py.

The variables which need to be modified for changing functionality:

• checkSum.py:120:: filterword # filters used to block
• firewall_monitor.py:28:: dns_servers # list of dns servers
• firewall_monitor.py:30:: allow_subnet # subnets to allow
• firewall_monitor.py:32:: reject_regex # subnets to reject

For simulation of the network topology:

$ sudo mn --topo=tree,1,3 --mac --controller=remote --switch ovsk,protocols=OpenFlow13 --nat

To run the firewall_monitor.py from ryu:

$ ryu-manager --verbose firewall_monitor.py

About

This project is a partial fullfillment for CS3543 (Networks II) offered by Dr.Kotaro Kataoka at IIT Hyderabad in Spring'20 semester.