Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

A QoE-Driven Personalized Incentive Mechanism Design for AIGC Services in Resource-Constrained Edge Networks

Hongjia Wu, Minrui Xu, Zehui Xiong, Lin Gao, Haoyuan Pan, Dusit Niyato, Tse-Tin Chan

TL;DR

A dual-perturbation reward optimization algorithm is developed, reducing the implementation complexity of adaptive pricing and incentivizing ASPs to provide personalized AIGC services under MEC resource constraints through a QoE-driven incentive mechanism.

Abstract

With rapid advancements in large language models (LLMs), AI-generated content (AIGC) has emerged as a key driver of technological innovation and economic transformation. Personalizing AIGC services to meet individual user demands is essential but challenging for AIGC service providers (ASPs) due to the subjective and complex demands of mobile users (MUs), as well as the computational and communication resource constraints faced by ASPs. To tackle these challenges, we first develop a novel multi-dimensional quality-of-experience (QoE) metric. This metric comprehensively evaluates AIGC services by integrating accuracy, token count, and timeliness. We focus on a mobile edge computing (MEC)-enabled AIGC network, consisting of multiple ASPs deploying differentiated AIGC models on edge servers and multiple MUs with heterogeneous QoE requirements requesting AIGC services from ASPs. To incentivize ASPs to provide personalized AIGC services under MEC resource constraints, we propose a QoE-driven incentive mechanism. We formulate the problem as an equilibrium problem with equilibrium constraints (EPEC), where MUs as leaders determine rewards, while ASPs as followers optimize resource allocation. To solve this, we develop a dual-perturbation reward optimization algorithm, reducing the implementation complexity of adaptive pricing. Experimental results demonstrate that our proposed mechanism achieves a reduction of approximately $64.9\%$ in average computational and communication overhead, while the average service cost for MUs and the resource consumption of ASPs decrease by $66.5\%$ and $76.8\%$, respectively, compared to state-of-the-art benchmarks.

A QoE-Driven Personalized Incentive Mechanism Design for AIGC Services in Resource-Constrained Edge Networks

TL;DR

A dual-perturbation reward optimization algorithm is developed, reducing the implementation complexity of adaptive pricing and incentivizing ASPs to provide personalized AIGC services under MEC resource constraints through a QoE-driven incentive mechanism.

Abstract

With rapid advancements in large language models (LLMs), AI-generated content (AIGC) has emerged as a key driver of technological innovation and economic transformation. Personalizing AIGC services to meet individual user demands is essential but challenging for AIGC service providers (ASPs) due to the subjective and complex demands of mobile users (MUs), as well as the computational and communication resource constraints faced by ASPs. To tackle these challenges, we first develop a novel multi-dimensional quality-of-experience (QoE) metric. This metric comprehensively evaluates AIGC services by integrating accuracy, token count, and timeliness. We focus on a mobile edge computing (MEC)-enabled AIGC network, consisting of multiple ASPs deploying differentiated AIGC models on edge servers and multiple MUs with heterogeneous QoE requirements requesting AIGC services from ASPs. To incentivize ASPs to provide personalized AIGC services under MEC resource constraints, we propose a QoE-driven incentive mechanism. We formulate the problem as an equilibrium problem with equilibrium constraints (EPEC), where MUs as leaders determine rewards, while ASPs as followers optimize resource allocation. To solve this, we develop a dual-perturbation reward optimization algorithm, reducing the implementation complexity of adaptive pricing. Experimental results demonstrate that our proposed mechanism achieves a reduction of approximately in average computational and communication overhead, while the average service cost for MUs and the resource consumption of ASPs decrease by and , respectively, compared to state-of-the-art benchmarks.

Paper Structure

This paper contains 32 sections, 5 theorems, 45 equations, 8 figures, 5 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Background and Motivation
  3. Solution and Contributions
  4. Related Work
  5. AIGC Services in Networks
  6. Multimodal Prompt Engineering
  7. Incentive Mechanisms for AIGC Services
  8. System Overview
  9. System Model
  10. Ambiguity of Language
  11. QoE of AIGC Services
  12. Problem Formulation
  13. Utility of ASPs
  14. Utility of MUs
  15. Equilibrium Problem with Equilibrium Constraints
  16. ...and 17 more sections

Key Result

Theorem 1

(The upper bound of the performance gap) When ASP $n$ infers the task of MU $m$ based on CoT prompting technology, consider a set of $K$ CoT examples $O_{k}^{nm} = (o_{k,r}^{nm})_{0 \leq r \leq l_k}$, generated from the intention $(I^{nm})^{*}$ with the optimal context $c_{n}^{*} \sim q(c_{n})$. The where $a^{nm}_{0}$, sampled from $q(\cdot|(I^{nm}_{0})^{*})$, is the input message or task generate

Figures (8)

  • Figure 1: Illustration of the proposed QoE-driven incentive mechanism in an MEC-enabled AIGC network. The network comprises multiple ASPs deploying differentiated AIGC models on edge servers, while multiple MUs with heterogeneous QoE requirements request services from these ASPs.
  • Figure 2: The workflow of the QoE-driven incentive mechanism. Each MU has distinct service requirements reflecting individual preferences. For example, vehicles demand concise and accurate responses to ensure safety, whereas content creators require richer inputs and outputs to support content generation.
  • Figure 3:
  • Figure 4: Example of different reasoning step lengths in CoT. When the number of CoT examples is fixed, using longer reasoning steps within each example generally improves accuracy. ASPs can therefore select CoT examples with varying step lengths to better approach the final answer.
  • Figure 5: The impact of three parameters on the decisions of MU 1 and ASP 1.
  • ...and 3 more figures

Theorems & Definitions (12)

  • Definition 1
  • Theorem 1
  • Lemma 1
  • proof
  • Definition 2
  • Definition 3
  • Theorem 2
  • proof
  • Theorem 3
  • proof
  • ...and 2 more