Interpolation-Based Optimization for Enforcing ℓp-Norm Metric Differential Privacy in Continuous and Fine-Grained Domains

This artifact provides the full MATLAB implementation and experimental pipeline for the anchor-interpolated privacy optimization (AIPO) framework, which enforces ℓp-norm Metric Differential Privacy (mDP) on continuous or fine-grained domains. It includes the following components: Experiment driver scripts (main_2norm.m, main_1norm.m, main_granularity.m, main_ablation_privacybudget.m): Each script reproduces a specific subset of tables/figures in the paper and writes outputs to the ./results/ subdirectories (e.g., cost/, violation/, time/, ppr/). Specifically: main_2norm.m produces Tables 2–4 and Figures 8–10 (Appendix); main_1norm_appendix.m produces Tables 8–10 (Appendix); main_granularity_appendix.m produces Figure 14 (Appendix); and main_privacyablation_budget.m produces Table 7 and Figures 11–13 (Appendix). Core implementation and helper functions (./functions/): This directory contains the optimization, interpolation, and evaluation routines implementing AIPO, as well as baseline methods (Exponential Mechanism, Laplace, TEM, COPT, LP, etc.). Datasets (./datasets/): This directory contains the processed OpenStreetMap road-network datasets (nodes and edges) for Rome, London, and New York City used in all experiments. Intermediate and result artifacts (./intermediate/, ./results/): ./intermediate/ caches reusable intermediate artifacts to avoid recomputation, while ./results/ stores the final metrics (utility loss, mDP violation ratio, runtime, PPR) for each mechanism and city.