Reducing Client-Side Cost in Secure Inner Product Evaluations

In the era of ubiquitous data collection and analysis, preserving privacy while utilizing data, such as inner product evaluations, poses a significant challenge. One such method is inner product functional encryption (IPFE), which enables the computation of inner products on encrypted vectors without exposing the vectors themselves. However, the computational intensity of IPFE decryption poses challenges, particularly for resource-limited client devices and high-dimensional vectors. To facilitate partial decryption, prior IPFE schemes require the client to derive a partial decryption key from an assigned secret key, incurring computation costs linear to the vector dimension. In this paper, we provide a framework for the challenges of IPFE, enabling efficient decryption outsourcing, maintaining data privacy, and facilitating key generation outsourcing in IPFE schemes. In the proposed framework, outsourced decryption using a partial decryption key generates a masked plaintext that only the client can unmask. Furthermore, the process of deriving the partial decryption key can be fully outsourced, thereby ensuring that the client's computational burden remains constant.