Nucleic acid-based cryptographic approaches are an innovative emerging field for information process. However, the poor reproducibility and interference from bioenvironment of the existing decryption led to different binary translation according to the fixed threshold defined by "Sender", which seriously affects the authenticity during message communication. Here, a programmed DNA constitutional dynamic network (CDN)-derived adaptive threshold is shown, which is defined by the difference value of the two groups of the output patterns from CDN. Under external stimuli, the threshold is adaptive to the generated dynamic output patterns, which avoids the contrary binary translation from a slight difference on the output under fixed threshold. Importantly, there are two self-calibrating patterns in each output group and the total concentration of constituents from the CDN system are constant, which greatly eliminates the data error. The CDN system is accompanied by computational simulation, which can predict the output patterns of the system at different states. The CDN is used to control the orthogonal and cascaded nanoparticle-based molecular amplifiers to expand the volume of the transmitting message, as well as allow the accurate and specific sensing of DNA. Various state-of-the-art representation is demonstrated by coding and decoding different types of messages.