Lightweight Reversible Watermarking for Real-Time Authentication on Resource-Constrained IoMT Edge Devices

The rapid proliferation of the Internet of Medical Things (IoMT) has shifted diagnostic data processing from centralized cloud servers to edge devices. While this reduces latency, these edge devices—often wearables or embedded sensors—are severely constrained in terms of power, memory, and clock speed. Ensuring the integrity and authenticity of medical data at the point of capture is critical; however, traditional cryptographic authentication methods introduce significant computational overhead and permanently alter the cover media. This paper proposes a novel Lightweight Reversible Watermarking (LRW) scheme specifically architected for real-time authentication on resource-constrained IoMT edge devices. Unlike conventional reversible watermarking techniques that prioritize capacity or visual quality at the expense of complexity, our algorithm optimizes the embedding-extraction cycle for ultra-low power consumption and minimal memory footprint. By utilizing a modified Difference Expansion (DE) technique paired with efficient histogram shifting, we achieve high tamper detection accuracy while maintaining data integrity. Experimental results on ARM Cortex-M0+ and RISC-V based platforms demonstrate that our method reduces energy consumption per bit by 62% compared to existing reversible schemes and completes authentication in under 15ms, enabling viable real-time operation. The watermarked medical images (X-ray, ECG, and fundus) are fully recoverable without information loss, satisfying strict regulatory requirements for diagnostic integrity.