Detection of smart card attacks using neural networks

The classification of the smartcard attacks is part of learning process used for understanding the different ways of breaking the system security. In fact, these classification methods can be applied for improving the system security, especially the smartcard one. In the paper, we study one of the smart classification methods, able to learn from data patterns and classify the attacks intelligently in order to respond as an appropriate case previously identified. According to this methodology, we develop first a neural network via the simulator JavaNNS, then we create a training database which contains attacks and class's models that the network must learn and then it must be able to classify other models contained in the test database and finally install the network generated in a javacard in order to get a smart one.