ResSen: Imager Privacy Enhancement Through Residue Arithmetic Processing in Sensors
Nedasadat Taheri, Sepehr Tabrizchi, Deniz Najafi, Shaahin Angizi, Arman Roohi · 2024
The increasing use of image sensors across various domains poses notable privacy challenges. In response, this paper introduces a novel architecture, namely ResSen, to enhance the privacy and efficiency of traditional image sensors. Our approach integrates the Residue Number System (RNS) with in-sensor digital encryption techniques to forge a robust, dual-layer encryption mechanism. By embedding RNS within analog-to-digital converters (ADCs), we significantly strengthen privacy measures, effectively countering different violations and ensuring the integrity and confidentiality of data transmissions. A key feature of our system is its programmable key, which complicates unauthorized output prediction or replication, providing a supe-rior encryption methodology. Notably, ResSen demonstrates that deactivating one of the moduli results in 25 % bandwidth savings at the cost of minor accuracy degradation. This underscores the practicality and effectiveness of our sensor architecture in addressing the dual objectives of privacy enhancement and operational efficiency.