Optimization for Semantic-Aware Resource Allocation under CPT-based Utilities

TL;DR

This paper addresses resource allocation in semantic communication by modeling agents' risk perceptions with CPT, formulating a complex optimization problem, and proposing an efficient algorithmic solution.

Contribution

It introduces a novel optimization framework combining SCA, subgradient, and Lagrangian methods for CPT-based utility maximization in semantic resource allocation.

Findings

The proposed algorithm effectively solves nonconvex, nonsmooth problems.

The framework balances computational efficiency with solution quality.

It demonstrates practical applicability in semantic-aware communication systems.

Abstract

The problem of resource allocation in goal-oriented semantic communication with semantic-aware utilities and subjective risk perception is studied here. By linking information importance to risk aversion, we model agent behavior using Cumulative Prospect Theory (CPT), which incorporates risk-sensitive utility functions and nonlinear transformations of distributions, reflecting subjective perceptions of gains and losses. The objective is to maximize the aggregate utility across multiple CPT-modeled agents, which leads to a nonconvex, nonsmooth optimization problem. To efficiently solve this challenging problem, we propose a new algorithmic framework that combines successive convex approximation (SCA) with the projected subgradient method and Lagrangian relaxation, Our approach enables tractable optimization while preserving solution quality, offering both theoretical rigor and practical effectiveness in semantics-aware resource allocation.