An IoT Encryption Communication Method Based on Neural Network

Secret communication is crucial for the Internet of Things (IoT) in certain spectral areas, such as smart grid. In this paper, a novel IoT encryption communication method based on neural network is proposed. The method employs a series of matrices to represent the ASCII characters in the plaintext. And then the random noise is further added to the matrix, generating the ciphertext. A neural network is built and trained to build up the connection between the characters and the matrices, which is utilized to decrypt the ciphertext. An experiment was made, in which the accuracy, bit error rate, and NIST statistical test results were evaluated. The results demonstrate that our method exhibits robust resistance against the impact of channel noise on both the bit error rate (BER) and the accuracy. Furthermore, the P-Value obtained in the NIST digital quality tests exceeds 0.01, providing evidence for the security of our proposed method.