Abstract:We consider a private discrete distribution estimation problem with one-bit communication constraint. The privacy constraints are imposed with respect to the local differential privacy and the maximal leakage. The estimation error is quantified by the worst-case mean squared error. We completely characterize the first-order asymptotics of this privacy-utility trade-off under the one-bit communication constraint for both types of privacy constraints by using ideas from local asymptotic normality and the resolution of a block design mechanism. These results demonstrate the optimal dependence of the privacy-utility trade-off under the one-bit communication constraint in terms of the parameters of the privacy constraint and the size of the alphabet of the discrete distribution.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is how to optimally estimate discrete distributions under the dual constraints of limited communication resources (specifically, a one - bit communication constraint) and privacy protection. Specifically, the researchers explored this problem under two privacy - protection mechanisms - Local Differential Privacy (LDP) and Maximal Leakage (ML) - and used the worst - case Mean Squared Error (MSE) as the evaluation criterion. ### Specific description of the problem 1. **Privacy protection**: Two privacy - protection mechanisms are considered in the study: - **Local Differential Privacy (LDP)**: Ensure that each individual's data does not leak too much private information when published. - **Maximal Leakage (ML)**: Limit the amount of privacy information leakage in an average sense. 2. **Communication constraint**: Each user can only send one - bit information (i.e., 0 or 1), which is the minimum non - trivial communication cost. 3. **Estimation error**: The worst - case Mean Squared Error (MSE) is used to measure the estimation error. ### Research objectives The goal of the paper is to fully characterize the trade - off relationship among privacy - utility - communication (Privacy - Utility - Communication Trade - off, PUCT) under the above conditions. Specifically, the researchers hope to find the optimal discrete distribution estimation method under given privacy protection and communication constraints and quantify its performance. ### Main contributions - **Theoretical results**: The paper provides exact asymptotic expressions of PUCT under the two privacy - protection mechanisms of LDP and ML. These results show the influence of privacy - protection parameters (such as \(\epsilon\), \(\delta\), \(\gamma\)) and the size \(v\) of the discrete distribution alphabet on PUCT. - **Optimal scheme design**: Proposed optimal estimation schemes based on the Block Design Mechanism, which can achieve optimal estimation performance while satisfying privacy and communication constraints. ### Formula summary - For the LDP mechanism, the PUCT expression is: \[ \text{PUT}_{\text{LDP}}(v, \epsilon, \delta) = \begin{cases} \frac{(v - 1)^2}{v} \left( \frac{e^\epsilon+ 1}{e^\epsilon+ 2\delta - 1} \right)^2 & \text{if } v \text{ is even and } \epsilon \geq \zeta(v, \delta), \\ \frac{(v - 1)^2}{v} \cdot \frac{(e^\epsilon+ 1)^2 + 4}{(e^\epsilon+ 2\delta - 1)^2} & \text{if } v \text{ is odd and } \epsilon \geq \zeta(v, \delta), \\ \frac{(v - 1)(v - \delta)}{v\delta} & \text{otherwise}. \end{cases} \] - For the ML mechanism, the PUCT expression is: \[ \text{PUT}_{\text{ML}}(v, \gamma) = \frac{(v - 1)(v - e^\gamma+ 1)}{v(e^\gamma - 1)} \] where, \[ \zeta(v, \delta) = \log \left( 1 + \frac{2 (\sqrt{\delta(v^* - 1)(v^* - \delta)} - \delta)}{v^*} \right) \] and \(v^*\) =

Optimal Private Discrete Distribution Estimation with One-bit Communication

Exactly Optimal and Communication-Efficient Private Estimation via Block Designs

Achieving the Exactly Optimal Privacy-Utility Trade-Off With Low Communication Cost via Shared Randomness

Optimal Private and Communication Constraint Distributed Goodness-of-Fit Testing for Discrete Distributions in the Large Sample Regime

Mutual Information Optimally Local Private Discrete Distribution Estimation

Simple Binary Hypothesis Testing under Local Differential Privacy and Communication Constraints

Instance-Optimal Differentially Private Estimation

Asymptotically optimal private estimation under mean square loss

On the Statistical Complexity of Estimation and Testing under Privacy Constraints

Differentially Private Distributed Nonconvex Stochastic Optimization with Quantized Communication

Instance-optimal Mean Estimation Under Differential Privacy

Strong Privacy and Utility Guarantee: Over-the-Air Statistical Estimation

Compressive Sensing Approaches for Sparse Distribution Estimation Under Local Privacy

Mean Estimation of Numerical Data Under (ε,δ) -Utility-Optimized Local Differential Privacy

On Communication Protocols that Compute Almost Privately

Optimal Estimation for Discrete-Time Linear System with Communication Constraints and Measurement Quantization

Mean Estimation of Numerical Data under (ϵ, Δ)-Utility-optimized Local Differential Privacy

Compressive Sensing Approaches for Sparse Distribution Estimation under Local Privacy

Minimax Rates of Estimating Approximate Differential Privacy

The Cost of Privacy: Optimal Rates of Convergence for Parameter Estimation with Differential Privacy