You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Server precomputes one global hint from the public matrix + DB; the same hint serves every client (client-independent), downloaded server→client only (one-directional).
10 GB/s throughput via plain LWE matrix-vector multiply
Group 2b — Client-specific hint (interactive)
Each client derives its own private hint via client↔server preprocessing (streaming the full DB, or a shared-key/FHE setup). Only category achieving sublinear online server time.
Comparison of SealPIR, MulPIR, Gentry-Ramzan for keyword PIR
Footnotes
The "MulPIR" referenced in KeywordPIR is the Ali et al. [1] implementation of Gentry & Halevi's "Compressible FHE with Applications to PIR" [2], which is already covered under mulpir_2019. Open-source implementations: [3] [4]. ↩