Research & Academic Publications

An analysis of security limitations and semantic notions bridging the gap between CPA and CCA2 security in fully homomorphic encryption (FHE). This report critiques the verified Chosen Ciphertext Attack (VCCA) frame proposed by Manulis and Nguyen, highlighting its theoretical robustness against underlying security models.

Data publication can greatly benefit private and public entities. This, however, shouldn’t be done at a great cost to the privacy of individuals whose data is published. In order to protect the anonymity of individuals in a data set, the propriety of k-anonymity was developed for data sets, giving a strong privacy guaranty. This paper will look into the computational complexity of the optimal solution, as well as the complexity of different proposed approximate algorithms.

Proposes "Data Over Light Onboarding" (DOLO), a novel onboarding protocol for smart smoke detectors utilizing ambient light sensors and standard smartphone camera flashes. This research framework replaces expensive and vulnerable SoftAP setups with secure, physically contained symmetric credential transfers.

A comprehensive survey detailing mechanisms to identify and authenticate users continuously via localized IoT sensors. The study compares machine learning model variations (e.g., Random Forests, SVMs, and RNNs) analyzing continuous biometrics such as gait patterns, voice recognition, and touch gesture behavior.

An analysis and critique of Surakav, a generative adversarial network (GAN) based trace generator designed to defend against advanced website fingerprinting (WF) attacks on the Tor network. The report evaluates its mechanics, performance overhead, and long-term implications for anonymous communication.