Cryptography is knowledge of manipulating data to conceal secure information. It serves an essential functionality in wide variety of applications. So, several encoding and decoding methods have been proposed to enhance cryptography techniques. In this study a novel approach based on multi layer cellular automata is proposed to be used in cryptography applications. Proposed multi layers cellular automata employs interaction between two heterogeneous cellular automata to imitate Pseudo-Neumann neighborhood structure and generate trackable random integers. These random numbers are assumed as time variant keys for encoding and decoding purposes. To verify and validate performance of proposed architecture, several simulations are performed. Simulation results prove that two-layer cellular automata generate more uniform random numbers in comparison with MATLAB. Consequently, proposed architecture demonstrated desirable behavior and has less risk. Furthermore, the architecture is suitable for hardware implementations.