KSOFM Network based neural key generation for wireless communication

Abstract

In this paper a single layer perceptron (KSOFM Network) based neural key generation for encryption/decryption for wireless communication has been proposed. Identical KSOFM network has been used in both sender and receiver side and the final output weight has been used as a secret key for encryption and decryption. The final output matrix of the KSOFM network is taken and the minimum value neuron in the matrix is considered as the secret key. Depending upon the input and output neuron, different keys has been generated for each session which helps to form secret session key. In sender side the plain text is encrypted with the secret key to form the cipher text done by the EX-OR operation between the secret key and the plain text. Receiver will use the same secret key to decrypt the cipher text to plane text. The secret key is not sent via any medium so it minimizes the man-in-the-middle attack. Moreover various tests are performed in terms of chi-square test, which shows comparable results with the said proposed system.