Research Projects

Today’s cyberspace is increasingly governed by applications doing several complex, resource-intensive computations on large-scale servers. They span across different domains, e.g., healthcare, social media, finance etc. Practically deployed technologies like cloud computing on big-data, artificial intelligence (AI) and machine learning (ML) have greatly boosted such developments through large-scale data storage, sharing and outsourcing computations on third-party cloud service providers. All these undoubtedly will continue to be a boon for progressing science, technology and humanity. However, sharing (sensitive) data in the clear transfer clients’ trust on such cloud severs. The heterogeneity of such applications makes it difficult to retain trust with these third parties responsible for computing on the collected data. This raises pressing issues on using such systems in practice. E.g., today fully scaled AI and ML applications work with enormous volumes of (potentially users’ personal) data. Many other applications have witnessed an upsurge of cyber breaches raising serious security and privacy concerns. With an ever growing number of applications, they also bring in critical risks with them like privacy and security of users and data. Furthermore, most cryptography actually used today base on hard problems (on classical computers) that are known to easily break on powerful quantum computers. Recently, tremendous efforts are being put in building real, scaled up quantum computers. This will again greatly boost current computing abilities, but it also urgently calls to build cryptography secure against quantum attackers. Hence, designing (possibly post-quantum) secure protocols, tailored for the above applications, is paramount now.

This project aims to build advanced (and possibly optimized) cryptographic tools that allow fine-grained processing on encrypted data. Some of it also aim to provide solutions tailored for applications. To this end, we will largely investigate useful extensions of functional encryption (FE). FE is an advanced form of public key encryption that allows to learn authorized functions of encrypted data without decrypting the entire data. The project aims to build and/or implement new FE systems with desirable features and useful applications in practice. These include computing on encrypted data of arbitrary size possibly originating from multiple, independent sources and/or minimizing the root of trust in FE. Further, the project also aims to look into applications of FE to build useful protocols, e.g., functional adaptor signatures (FAS) for rich function classes. FAS enable partial sale of secret data held by sellers to authorized buyers while ensuring privacy of the seller’s database. This makes FAS useful to sell data (possibly over blockchains) where the database may have sensitive information. The project also aims to build (& possibly benchmark) the tools discussed above from lattice-based post-quantum security.