Differentially private data aggregating with relative error constraint

Abstract Privacy preserving methods supporting for data aggregating have attracted the attention of researchers in multidisciplinary fields. Among advanced methods, differential privacy (DP) has become an influential mechanism owing to its rigorous guarantee and high utility. But DP no limitation on bound noise, leading a low-level Recently, investigate how while limiting relative error fixed bound. However, these schemes destroy statistical properties, including mean, variance MSE, which are foundational elements analyzing. In this paper, we explore optimal solution, novel definitions implementing mechanisms, maintain properties satisfying with Experimental evaluation demonstrates that our outperforms current terms security utility large quantities queries.