Hybrid RBFN Based Encrypted SSH Traffic Classification

This paper presents a hybrid Radial Basis Function Network (RBFN) based machine learning approach for encrypted data traffic classification. We concentrate on classifying the SSH-based encrypted traffic from the non SSH traffic. The hybrid RBFN network is employed with k-means clustering and gradient descent learning technique for the binary classification. The suggested technique is experimented on standard datasets including AMP, MAWI, DARPAA99, and NIMS with k-fold cross-validation. Simulation results show that RBFN network is able to provide flow accuracy of more than 99% for all the datasets. Comparative analysis is made to show the efficacy of the proposed scheme with other competent techniques.