High Speed Pseudo Random Number Generator for Homomorphic Encryption

As digital services offered through the cloud have proliferated over the last decade, ensuring the privacy of user data has become increasingly critical. To preserve data encryption through all activities carried out during service usage, Fully Homomorphic Encryption (FHE) offers a secure solution. A Pseudo-Random Number Generator (PRNG) is an essential part of this encryption process that is highly impactful on the security and effectiveness of the system. In this work, we propose a novel high-speed pseudo-random number generator design based on the ChaCha stream cipher. By incorporating a Linear Feedback Shift Register (LFSR)-based counter and decreasing the number of rounds in the ChaCha algorithm, the proposed pseudo-random number generator achieves a balance between both security and performance. With increased efficiency both in terms of speed and area utilization, the design operates at a clock frequency of 277.78 MHz while delivering a throughput of 2777.8 Mbps. It successfully passes all the NIST statistical tests, demonstrating strong randomness properties that are crucial for cryptographic applications. Compared to prior PRNG designs ([5], [8], and [9]), our ChaCha6 PRNG achieves a 30 % higher throughput while maintaining strong statistical randomness.