Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Multi-Task Private Semantic Communication

Amirreza Zamani, Sajad Daei, Tobias J. Oechtering, Mikael Skoglund

TL;DR

The paper tackles private semantic communication under a privacy leakage constraint in a multi-task setting. It models the disclosed data as $U=f(X)+M$, balancing privacy $I(S;U)\le\epsilon$ with goal-oriented utility $I(U;h_i(X))$, and extends from a single task to multiple prioritized tasks. Using extended FRL/SFRL constructions and a separation technique, it derives both lower and upper bounds for the single-task trade-off, and then proves that the multi-task problem decomposes into parallel single-task problems, enabling a simple, scalable noise-design approach. The results provide explicit bounds and a constructive pathway to design privacy mechanisms that are effective across tasks while guaranteeing privacy, with potential impact on efficiency and security in goal-oriented semantic communication systems.

Abstract

We study a multi-task private semantic communication problem, in which an encoder has access to an information source arbitrarily correlated with some latent private data. A user has $L$ tasks with priorities. The encoder designs a message to be revealed which is called the semantic of the information source. Due to the privacy constraints the semantic can not be disclosed directly and the encoder adds noise to produce disclosed data. The goal is to design the disclosed data that maximizes the weighted sum of the utilities achieved by the user while satisfying a privacy constraint on the private data. In this work, we first consider a single-task scenario and design the added noise utilizing various methods including the extended versions of the Functional Representation Lemma, Strong Functional Representation Lemma, and separation technique. We then study the multi-task scenario and derive a simple design of the source semantics. We show that in the multi-task scenario the main problem can be divided into multiple parallel single-task problems.

Multi-Task Private Semantic Communication

TL;DR

The paper tackles private semantic communication under a privacy leakage constraint in a multi-task setting. It models the disclosed data as , balancing privacy with goal-oriented utility , and extends from a single task to multiple prioritized tasks. Using extended FRL/SFRL constructions and a separation technique, it derives both lower and upper bounds for the single-task trade-off, and then proves that the multi-task problem decomposes into parallel single-task problems, enabling a simple, scalable noise-design approach. The results provide explicit bounds and a constructive pathway to design privacy mechanisms that are effective across tasks while guaranteeing privacy, with potential impact on efficiency and security in goal-oriented semantic communication systems.

Abstract

We study a multi-task private semantic communication problem, in which an encoder has access to an information source arbitrarily correlated with some latent private data. A user has tasks with priorities. The encoder designs a message to be revealed which is called the semantic of the information source. Due to the privacy constraints the semantic can not be disclosed directly and the encoder adds noise to produce disclosed data. The goal is to design the disclosed data that maximizes the weighted sum of the utilities achieved by the user while satisfying a privacy constraint on the private data. In this work, we first consider a single-task scenario and design the added noise utilizing various methods including the extended versions of the Functional Representation Lemma, Strong Functional Representation Lemma, and separation technique. We then study the multi-task scenario and derive a simple design of the source semantics. We show that in the multi-task scenario the main problem can be divided into multiple parallel single-task problems.
Paper Structure (3 sections, 4 theorems, 29 equations, 2 figures)

This paper contains 3 sections, 4 theorems, 29 equations, 2 figures.

Table of Contents

  1. Introduction
  2. system model and Problem Formulation
  3. Main Results

Key Result

Theorem 1

For any $0\leq \epsilon< I(S;h(X))$ and joint distribution $P_{S,f(X),h(X)}$, we have where with $\alpha=\frac{\epsilon}{H(S)}$ and $\alpha_2=\frac{\epsilon}{H(S_2)}$ for any representation $S=(S_1,S_2)$. The lower bound is tight if $H(S|h(X))=0$, i.e., $S$ is a deterministic function of $h(X)$. Furthermore, if the lower bound $L_{h}^{1}(\epsilon)$ is tight then we have $H(S|h(X))=0$.

Figures (2)

  • Figure 1: Single-task private semantic communication model. The goal is to design disclosed data $U$ such that it keeps as much information as possible about the task $h(X)$ while satisfying a certain privacy constraint.
  • Figure 2: Multi-task private semantic communication model. The goal is to design $U$ that maximizes weighted linear combination of utilities while satisfying a certain privacy constraint.

Theorems & Definitions (8)

  • Remark 1
  • Remark 2
  • Theorem 1
  • Example 1
  • Theorem 2
  • Theorem 3
  • Theorem 4
  • Remark 3