Study Of The Mem-Elements-Based Cellular Neural Network

Since the T_iO₂-film practical memristor device was fabricated by HP Labs in 2008, there have been widespread attention on memristor, memcapacitor and meminductor, and the study on mem-elements-based neural network system has made quickly progress. The cellular neural network is a kind of nonlinear signal processing system, combining mem-elements with the cellular neural network is promising in achieving better circuit performance.In this thesis, several modified schemes of the cellular neural network based on the properties of memristor and memcapacitor are proposed, the corresponding mathematical analyses and simulations of applications are also carried out. The main contents are as follows: firstly, an improvement in circuit of the synaptic weights is proposed. In this scheme, the input AC signals are separated from the memristor-weights, at the same time, the communicating lines among the neighbor cells can be reduced. Secondly, improvements with memristors and memcapacitors in cell circuits are proposed. A memcapacitor based cellular neural network cell is presented, together with analysis of the time stability of the new system, and simulation of its image processing ability. A trilaminar memristor-cell is proposed. By means of drawing the phase portraits, bifurcation diagrams and the Lyapunov exponent spectrums, it is found that the system is chaotic under the condition of certain parameter ranges. With appropriate parameters, a set of double helical chaotic signals can be generated. Let this set of chaotic signals be the key source, we got the scrambling matrix and implemented simulation of a grayscale image's encryption and decryption, numerical analyses of the simulation result are also presented.The above work provides some new ideas for study of improving cellular neural network circuits. They can be helpful to make the cellular neural network benefit from the excellent circuit characteristics of the mem-elements and achieve better performance as well as higher value of practical application.