Sherlock-Hou/Vulnerability-of-XOR-Arbiter-PUFs

Breaking the large XOR-Arbiter-PUFs using Neural Network

Vulnerability-of-XOR-Arbiter-PUFs

Breaking the large XOR-Arbiter-PUFs using Neural Network.

Objective

We will predict the final output respnse from k-XOR Arbiter PUFs with n-bit challenge input using keras.

Dataset Overview

We can access the dataset from here.

1. 5-XOR_128bit dataset: This dataset is generated using 5-XOR arbiters PUF with 128-bit stages or 128-bit challenge input. It consists of 6 million rows and 129 attributes where the last attribute is the class label or final response (1 or -1). It is divided into two sets: training set (5 million) and testing set (1 million).

2. 6-XOR_64bit dataset: This dataset is generated using 6-XOR arbiters PUFs with 64-bit stages or 64-bit challenge input. It consists of 2.4 million rows and 65 attributes where the last attribute is the class label or final response (1 or -1). It is divided into two sets: training set (2 million) and testing set (400K).

References

1. https://www.semanticscholar.org/paper/A-Machine-Learning-Based-Security-Vulnerability-on-Aseeri-Zhuang/c77f42238e6098000c7add21d517c7ff0676f54f

2. https://dl.acm.org/citation.cfm?id=1866335

3. https://machinelearningmastery.com/keras-functional-api-deep-learning/

4. https://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=6823677

5. https://www.appliedaicourse.com/