Distributed Information Encoding Using Self-Organized Patterns

Dynamical systems often generate distinct outputs according to different initial conditions, and one could infer the corresponding input configuration given an output. This property captures the essence of information encoding and decoding. Past studies have explored the possibility of encoding using chaos, where small changes in initial condition can lead to drastically different outputs. While the high sensitivity makes malicious attacks without prior knowledge of the system difficult, the unavoidable noise could make decoding challenging for the designated recipient. In contrast, biological self-organized patterns share global similarity but vary in detail due to random noise under the same or similar conditions, which could balance the trade-off between reliable decoding and security. Here, we demonstrated the use of self-organized bacterial colony patterns, combined with machine learning (ML), to achieve distributed information encoding and decoding with guaranteed security.

To encode, a message is converted into cell seeding configuration followed by colony growth, during which a colony pattern develops; to decode, we input the pattern into a trained CNN to convert it back to the original message. We propose two criteria for choosing suitable patterning systems: sufficient visual complexity and weakly chaotic mapping. Both are determined by the patterning dynamic and encoding setup. By modulating these factors, we could tune the trade-off among encoding capacity, decoding accuracy and security, characterized by ML decoding performance. We also implemented ensemble techniques for enhancing decoding reliability and making full use of the expensive-to-obtain patterning data. Moreover, our approach was combined with established cryptography techniques (e.g., encryption and hashing) to further enhance the security.