An efficient group synchronization of chaos-tuned neural networks for exchange of common secret key

For the design of the public-key exchange protocol, a chaos-based Triple-Layer Tree Parity Machine (TLTPM) directed group neural synchronization is presented in this study. For neural synchronization, a particular neural network structure called Tree Parity Machine (TPM) is utilized. Two TPMs take the same input but have distinct weight vectors, and update the weights using the neural learning algorithm by swapping output. It leads to perfect synchronization in certain phases, with the weights of the two TPMs becoming identical. The secret key is made up of these similar weights. However, there has not been much research on the randomness of the common input vector utilized in the synchronization process. A logistic chaotic system based on TLTPM is proposed in this paper. This proposed TLTPM model has several advantages like (1) leverages logistic chaos produced random common input vector for quicker synchronization. (2) With the same input, output, and hidden neurons, the suggested TLTPM model is quicker and more secure than TPM. (3) A binary tree framework considers each neural network as a node. (4) When the binary tree's i-th and j-th nodes are synchronized, one of these two nodes is chosen as the leader. (5) This leader node will now coordinate with the other branch's leader. After this procedure is completed, the coordinated weight has now become the cluster's session key. (6) And no need to synchronize each neural network in the group; rather, by coordinating between the elected leader nodes with just logarithmic synchronization steps, the whole group may share the same secret key. A variety of parametric tests have been performed on the suggested approach. The results were compared to some more contemporary methods. The suggested technique's findings have proven that it has both effective and resilient potential.