High-Density Chess Encryption: A CNN and DQN Framework for Steganography

This paper introduces a novel encryption framework that embeds digital information into sequences of legal chess moves. Traditional encryption schemes effectively render data unintelligible, but frequently fail to obfuscate the existence of the encrypted payload, thereby increasing its susceptibility to traffic analysis and targeted adversarial actions. Leveraging the complex and universally recognized structure of chess, we propose an innovative cryptographic approach that discretely encodes binary information within plausible chess games. The game of chess, with its vast, discrete state space and universally understood rule-set, presents a compelling yet challenging medium for steganography. Our contribution is a unique encoding algorithm that sequentially maps fixed-length binary segments to a unique, legal chess move. However, this direct mapping approach faces two inherent challenges: the high computational latency of engine-based move validation and a tendency to produce short, strategically naive games that are easily distinguished from human play, compromising covertness. To overcome these limitations, we engineered a multi-layered AI framework built upon the mapping algorithm. This hybrid system leverages a lightweight Convolutional Neural Network (CNN) and a Deep Q-Network (DQN) agent to navigate the complex trade-offs between data embedding, security, and strategic plausibility. Experimental results validate our approach, demonstrating that the complete AI-driven system produces significantly more stable and dense encodings than baseline methods. This work establishes a viable methodology for intelligent steganography within complex, symbolic environments.