A Novel Way to Generate Adversarial Network Traffic Samples against Network Traffic Classification

Algorithm 1

Adversarial samples of network traffic crafting algorithm.

Input:Normal Network Traffic

Output:Adversarial Samples of Network Traffic

BEGIN.

1.Preprocess (TF); //Pre-process and Extract characteristic ;

2.TranspPcapToIDX (TF); //Transform from pcap format to IDX format;

3.Normalized (); //Delaminate each characteristic dimension and normalize into section [0,255];

4.Reshape (TF); //Reshape each characteristic value of multiple types of characteristic as a grey value;

5.Visualization (TF); //Form a matrix and visualize the network traffic;

6.Training (TF, mode); //Train CNN models

7.Test (TF); //Test the accuracy of normal network traffic;

8.CraftingPerturbation (method); //use different methods of perturbation crafting to craft perturbation;

9.TA = GenerateAdvSample (); //, overlay the perturbation and original traffic to craft adversarial samples of network traffic

10.Visualization (TA); //Compare and results from Step 5

11.Evaluate (TA); //evaluate adversarial samples of traffic network being crafted

12.Return; //output adversarial samples of traffic network.

END