Fig 1.
The mirror two-layered NN used for encryption/decryption and signature.
Fig 2.
An architecture of recurrent and feedforward neural networks.
(a) Recurrent Neural Network, (b) Feedforward Neural Network.
Fig 3.
Training error versus epochs (an epoch is a whole batch of input vectors).
Table 1.
The recurrent NN performance.
Fig 4.
The encryption process using NN.
Fig 5.
The control flow of the distortion layer.
Fig 6.
The algorithmic steps of the random generator.
Fig 7.
The S-BOX.
Fig 8.
The S-BOX inverse.
Fig 9.
The dual diffusion method: The processing instructions and flow of control.
Fig 10.
The dual diffusion inverse method: The processing instructions and the flow of control.
Table 2.
The F-TAB: 14 bitwise-distortion actions.
Fig 11.
The key expansion process reproduced from [53].
Table 3.
Avalanche effect of dual diffusion method.
Table 4.
Key avalanche test.
Table 5.
Plaintext avalanche test.
Table 6.
Plaintext/Ciphertext correlation test.
Table 7.
Result of ENT test: Plaintext avalanche (Tp), Key avalanche (Tk), and Plaintext/Ciphertext correlation (Tc).
Table 8.
The proposed technique efficiency compared to other novel techniques (time in milliseconds).