Novel Extreme Multistable Tabu Learning Neuron: Circuit Implementation and Application to Cryptography

The complex dynamics of a simple memristive tabu learning neuron (MTLN) are considered in this article. The analysis of the stability of its equilibria revealed that it displays self-excited dynamics. The investigation of the dynamics of the considered model highlighted that it is extremely sensitive to the initial conditions. That sensitivity to the initial conditions is supported by the coexistence of an infinite number of stable states for the same set of system parameters but using different initial states. Among the infinity of coexisting stable states, there are periodic, quasiperiodic, and chaotic ones. The coexistence of an infinite number of chaotic attractors found in this work and not yet reported in such a model represents the first important contribution of this work. The circuit of the coupled neuron is also realized in the PSPICE simulation environment to further support the obtained result in extreme multistability. Therefore, the revealed chaotic dynamics of the MTLN is applied to compress and encrypt digital medical images. The compressed sensing (CS) approach is combined with deoxyribonucleic acid (DNA) coding/decoding to achieve high compression/encryption performances, including very low computational cost (encryption time
t
= 0.162 ms, encryption throughput ET = 1618.1 MB/s, number of CPU cycles NC = 1.8) useful for real-time compression. The compression/encryption method developed in this work represents the second main contribution based on the results of the analysis metrics.