Technology Innovation Institute (TII), the applied research pillar of Abu Dhabi’s Advanced Technology Research Council (ATRC), has announced that its Cryptography Research Centre (CRC) has launched the UAE’s first secure cloud technologies programme.
TII said that its secure cloud technologies programme aims to advance Privacy Enhancing Technologies (PETs), including fully homomorphic encryption (FHE), a form of encryption that permits users to perform computations on encrypted data without first decrypting it, and secure multi-party computation (MPC), creating methods for parties to jointly compute a function over their inputs while keeping those inputs private.
This line of research will be coupled with the field of verifiable computation for Machine Learning, to cater for a proof of correctness for an Inference-as-a-Service use case when data privacy is not required.
The research team in-charge is also joining efforts with the hardware cryptography team to develop FHE hardware accelerators to offload certain computing tasks onto specialised hardware components within the system, enabling greater efficiency.
Speaking on the announcement, Faisal Al Bannai, Secretary General of the Advanced Technology Research Council (ATRC), said: “Cloud computing has undergone significant growth in the last decade which has raised security and privacy challenges. Traditional approaches still require data to be decrypted for processing and a cloud-centralised key management system, thus exposing both the data and the secret key to cloud providers. This underscores the importance of launching this programme by our researchers at TII’s Cryptography Research Centre.”
Dr Najwa Aaraj, Chief Researcher at the Cryptography Research Centre, said: “FHE allows performing arbitrarily complex, dynamically chosen computations on data while it remains encrypted despite not having the secret decryption key. Moreover, FHE is a key enabler for multi-party computation (MPC) protocols implementing oblivious federated learning models, which are on the rise in critical infrastructure data transfer.”
“TII’s implementation of secret sharing-based MPC leverages FHE building blocks for providing active security and optimised preprocessing to speed up the oblivious implementation of Machine Learning models,” she said.
“Moreover FHE remains computationally intensive. Researchers at TII are also developing hardware accelerators driven by principal components and operations analysis and targeted enhancements using ASICs and multiple FPGA platforms. Further direction includes implementation on embedded processors with an extensible instruction set architecture.”