A key-controlled watermarking framework for tamper localization in physiological signals
Saifur Rahman, Iynkaran Natgunanathan, Chandan Kumar Karmakar · Computers in Biology and Medicine · 2025
Ensuring the authenticity and integrity of physiological signals is critical in modern healthcare, where electrocardiogram (ECG) data are frequently transmitted, stored, and analyzed across connected but potentially insecure platforms. This study proposes a lightweight and secure tamper detection framework for ECG signals based on digital watermarking. Patient-specific identifiers are embedded by modifying one of the least significant bits of each sample, while embedding positions are determined using a secret key to prevent unauthorized access and extraction. The approach preserves diagnostic quality, introduces no visible distortion, and allows precise localization of tampered regions. Experimental evaluation demonstrates that the framework achieves high robustness against a range of tampering conditions, including amplification, Gaussian noise, and re-equalization. Across all tested conditions, with secret key lengths of 6, 12, and 18 bits, the system consistently achieved 100% detection accuracy and F1 scores. These results confirm that the proposed method provides reliable and comprehensive tamper detection performance. The lightweight design enables efficient deployment in resource-constrained environments such as Internet of Things (IoT)-based healthcare devices. While validated on ECG signals, the framework is readily applicable to other physiological data, including electroencephalogram (EEG) and photoplethysmogram (PPG). To the best of our knowledge, this is the first tamper detection framework for physiological signals that simultaneously ensures perfect detection accuracy, preserves signal quality, supports low-complexity implementation, and incorporates cryptographic security.