A Dynamic AES Encryption Based on Memristive Chaos Neural Network

: This paper proposes an Advanced Encryption Standard (AES) encryption technique based on memristive neural network. A memristive chaotic neural network is constructed by the use of the nonlinear characteristics of the memristor. The chaotic sequence, which is sensitive to the initial value and has good random characteristics, is used as the initial key of AES grouping to realize "one-time-one-secret" dynamic encryption. Results show that the algorithm has higher security, larger key space and stronger robustness than the conventional AES. It can effectively resist the initial key fixed and exhaustive attacks.


Ⅰ. Introduction
Advanced Encryption Standard (AES), a group symmetric encryption with variable key length, take advantages of good security, high efficiency, easy implementation and strong flexibility, and has become an international mainstream standard encryption system. [1][2][3] However, there are still some security problems in AES, such as the fixed initial key, key decoding, and limited key space. [4][5][6][7] Chaotic systems were introduced to improve the AES encryption algorithm's security. [8][9][10][11][12] In 2004, A one-way coupled spatiotemporally chaotic map lattice was used to construct a new AES cryptosystem. 8 in 2018, a novel chaos-based hybrid encryption algorithm design for secure and effective image encryption was presented. 9 In 2019, an image encryption algorithm was proposed based on the combination of the chaos sequence and the modified AES. 10 In 2020, A four-dimensional chaotic system was applied to generate key and improve advanced encryption standard. 11 In 2021, a modified AES cryptosystem with dynamic random keys based on chaos synchronization was presented. 12 As the fourth fundamental circuit component, memristor has many advantages, 13 such as non-linearity, memory properties, low power consumption, simple structure, and so on. 14 There is also a lot of research on building chaotic systems and neural networks based on memristors. [15][16][17][18][19][20][21][22][23][24][25][26][27][28][29][30] For example, in 2008, it was derived that several nonlinear oscillators from Chua's oscillators by replacing Chua's diodes with memristors. 15 In 2012, delayed switching effect was used to control the switching of a memristor synapse between two neurons. 17 In 2018, I. Boybat present a multimemristive synaptic architecture with an efficient global counter-based arbitration scheme for a spiking neural network. 19 A simple and effective method was given for designing the autonomous memristor chaotic systems of infinite chaotic attractors. 21 In 2019, a chaotic memristor, which only contains meminductor, and memcapacitor models, was presented. 22 In 2020, a class of electro-optical memristors was introduced, which can emulate the key properties of synapses and neurons. 25 A physical memristor based on the Muthuswamy-Chua chaotic system (circuit) was provided. 27 In this paper, based on the memristor-based transient chaotic neural network (MTCNN), 31 the chaotic state for a long time is realized by the MTCNN. By MTCNN, AES initial key is dynamically generated to realize "one time one secret." Simultaneously, to improve the security, the Rivest-Shamir-Adleman(RSA) encryption is used to encrypt the initial parameters of the chaotic network. The sensitivity, key space, information entropy, correlation indexes, and poker test have been carried out and the capability and improvement of this proposed AES cryptosystem have been examined. As a chaotic neural network is sensitive to the initial value, with the slight modification of the initial value each time, a different non-duplicate key sequence can be obtained, conforming to the key standard. The key generated from each initial value can be used for one round of AES encryption, and ten initial values can complete one AES block encryption.

Ⅲ. Discussion
We examine the text encryption and decryption by the proposed

Ⅳ. Conclusion
In summary, this paper proposes an improved AES algorithm based on MTCNN. Based on the nonlinear characteristics of memristive chaotic neural network, the chaotic sequence, which has good random characteristics, is used as the initial key of AES grouping to realize "onetime-one-secret" dynamic encryption. The proposed AES algorithm can effectively improves the properties of seed key, key space, and anti-attack ability of AES algorithm.

Acknowledgments
This work is supported by NSFC under project Nos. 61774028, 92064004 and 61771097.

Data availability
The data that support the findings of this study are available from the corresponding author upon reasonable request.

Author information
State    x .