An Efficient Private Information Retrieval Scheme Based on Coefficient Extraction and Lattice-Based Cryptography

Private Information Retrieval (PIR) enables users to retrieve records from a database without revealing their query targets to the server. While lattice-based PIR schemes have gained attention due to their quantum-resistance properties, existing approaches face significant challenges in terms of computational overhead and noise management. Current state-of-the-art solutions like SealPIR, OnionPIR, and SPIRAL suffer from three main limitations: (1) complex ciphertext query expansion requiring iterative homomorphic operations,(2) substantial computational costs due to polynomial operations in ring-LWEbased schemes, and (3) inefficient conversion between heterogeneous ciphertexts that increases both communication overhead and noise growth. In this paper, we propose a novel coefficient extraction-based PIR scheme that addresses these challenges through two key innovations: A direct coefficient extraction technique that eliminates iterative homomorphic operations; a unified algebraic structure across RLWE, LWE, and LWEGSW schemes for seamless ciphertext conversion. Compared to existing approaches, our scheme achieves: significant reduction in computational overhead through optimized homomorphic operations compared to baseline approaches; substantially decreased communication costs via efficient ciphertext structure design- Better controlled noise growth allowing smaller parameter settings; maintained security based on standard RLWE/LWE hardness assumptions. These improvements make our scheme particularly suitable for practical PIR applications requiring both efficiency and strong security guarantees.