Digital University Kerala (left) Alex James and California University, Berkeley’s Leon Chua.
A collaborative study associated with Digital University Kerala has introduced an advanced digital watermarking technology that embedded the device-specific watermark in digital images, providing strong security against machine learning-based attacks.
The model promises applications to increase digital security in intellectual property security, safe image transmission and IOT device authentication.
The innovative method uses processing in the in-memory physical unaclaxy functions (PUFS) combined with the cellular memorister network (Cenns). Pufs produce unique digital fingerprints for objects depending on their physical properties, making them difficult to repeat them under different circumstances.
Researchers developed by Alex James (Digital University Kerala), Chithra Raghuvaran (University College Dublin) and Leon Chua (University of California, Burkeley), improves security against technology forgery and unauthorized imitation. Study, colleague reviewed in magazine IEEE transactions on emerging subjects in computing, Advanced machine highlights the ability of the system to oppose learning attacks.
Unique
The system works by creating a unique “fingerprint” for each image using random behavior of memesers, which are small electronic equipment. It makes anyone extremely difficult to copy the watermark with sophisticated software.
What sets this method apart from traditional techniques is its energy efficiency. Unlike the older methods relying on different processors, this technique directly processes information within memory, reduces power consumption and makes the watermark even more difficult to repeat.
Researchers say that potential applications for this technique are huge. This can revolutionize intellectual property protection by ensuring authenticity of digital images and preventing material fraud. The system promises safe image transmission in sensitive areas such as medical imaging, monitoring and military communication. It also provides a strong solution for device authentication in the IOT environment, which increases hardware-level security.
Despite their promising benefits, the researchers accepted the challenges in verifying watermark integrity when images are compressed, cropped or transformed post-ambed. Future research is expected to detect advanced techniques, such as fruitless watermarking and machine learning-based recovery to solve these issues.
Published – February 18, 2025 07:07 pm IST
