AI-Enhanced High-Dimensional Chaotic Encryption for Secure Satellite-Driven Smart Farming Networks

Abstract

Satellite-based sensing and IoT-based farm monitoring has become a key component of climate-smart agriculture to enhance crop productivity, soil health, and resource optimization. In an era where smart farming networks transmit high-resolution satellite and sensor data continuously across distributed networks, however, there are key security challenges, such as data tampering, interception, and unauthorized access. To overcome these challenges, this paper introduces the AI-Enhanced High-Dimensional Chaotic Encryption Framework tailored for secure smart farming ecosystems using satellites. Proposed multi-dimensional chaotic maps are combined with adaptive key modulation using AI techniques to produce more unpredictable and dynamic encryption keys. Lightweight Chaotic Permutation-Diffusion Cipher for remote sensing imagery and sensor telemetry data, providing confidentiality without excessive computational cost, appropriate for edge devices. Moreover, the dynamic adaptation of the encryption parameters with the use of a Reinforcement Learning (RL) module, based on the level of threats and the network load, improves the resilience against adaptive adversarial attacks. The experimental test on Sentinel-2, Landsat-8 and UAV multispectral datasets shows that the security performances are good with high NPCR (>99 %), UACI (>33 %), Shannon entropy (7.9998) and negligible transmission latency overhead. The framework enhances the data Integrity, privacy, and reliability, and allows scalable and secure AI-based precision agriculture within real-world smart farming networks.