Adaptive Network Monitoring for Online Classification of Tor/nonTor Traffic near to real-time.

Requirements:

• conda create --name rl-cnn python=3.7 --file requirements.txt
• conda activate rl-cnn
• sudo apt-get install libopenmpi-dev
• sudo apt-get install python3-scapy
• cd gym-basic
• python setup.py install You should proceed with this steps in each machine where you want to run. To register the environment.

➡️ Need to run dataset traffic before any task.

The dataset was build from CiC Univsersity of New Brunswich (Canadian) available here . This dataset contains all non-Tor classes such as audio, browsing, chat, email, p2p, transfer, video and voip. We merged all data of these classes in a single dataset named non-Tor.pcap. After, we used ISCXTor2016 as a Tor Traffic.

Run syntetic Traffic generator based on Tor/nonTor Dataset

./syntetic_packet_workload_gen.sh

➡️ Interface pooling test:

sudo python3 pooling.py 1 1 eth0

Where first parameter refers to amount of packet capture, second refers to time of capture and last one refers to interface to be monitored. Note that pooling command should run as root. Hence, install torch and its dependences as root accordingly

➡️ Finnaly, run: 🔛

 python3 teste.py --gamma 0.9 --env "gym_basic:basic-v1" --n-episode 200 --batch-size 1 --hidden-dim 12 --capacity 1000 --max-episode 50 --min-eps 0.01