HG‑LDP: A Memory‑Efficient Framework for High‑Frequency Item Statistics under Local Differential Privacy
This article introduces the HG‑LDP framework, which combines local differential privacy with the HeavyGuardian data structure to enable accurate, privacy‑preserving high‑frequency item statistics on streaming data while using only limited memory, and evaluates four algorithmic variants (BGR, DSR, BDR, CNR) through extensive experiments on synthetic and real‑world datasets.