Privacy-Aware Joint DNN Model Deployment and Partition Optimization for Delay-Efficient Collaborative Edge Inference

2025-02-22

AI Summary

This paper addresses the joint deployment optimization of deep neural networks (DNNs) for edge collaborative inference under resource constraints and privacy requirements, aiming to minimize long-term average inference latency. We propose the first integrated framework combining Lyapunov optimization, coalition game modeling, and greedy heuristics: Lyapunov optimization enables dynamic resource–latency trade-offs; coalition game theory jointly models server association and model partitioning decisions; and implicit differential privacy constraints enforce end-to-end privacy budget adherence. The framework simultaneously achieves low latency, strong privacy guarantees, and scalability in dynamic edge environments. Simulation results demonstrate a 23.6%–38.1% reduction in inference latency compared to baseline methods, strict compliance with the prescribed privacy budget, and robust performance under highly variable workloads.

Technology Category

• Machine Learning: Learning on the Edge & Model Compression
• Search and Optimization: Learning to Search
• Constraint Satisfaction and Optimization: Distributed CSP/Optimization

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Security and Privacy: Privacy-enhancing technologies
• Economics, Online Markets and Human Computation: Incentives in network design for Web infrastructures and ecosystems

Abstract

Edge inference (EI) is a key solution to address the growing challenges of delayed response times, limited scalability, and privacy concerns in cloud-based Deep Neural Network (DNN) inference. However, deploying DNN models on resource-constrained edge devices faces more severe challenges, such as model storage limitations, dynamic service requests, and privacy risks. This paper proposes a novel framework for privacy-aware joint DNN model deployment and partition optimization to minimize long-term average inference delay under resource and privacy constraints. Specifically, the problem is formulated as a complex optimization problem considering model deployment, user-server association, and model partition strategies. To handle the NP-hardness and future uncertainties, a Lyapunov-based approach is introduced to transform the long-term optimization into a single-time-slot problem, ensuring system performance. Additionally, a coalition formation game model is proposed for edge server association, and a greedy-based algorithm is developed for model deployment within each coalition to efficiently solve the problem. Extensive simulations show that the proposed algorithms effectively reduce inference delay while satisfying privacy constraints, outperforming baseline approaches in various scenarios.

Problem

• Optimize DNN model deployment
• Minimize inference delay
• Ensure privacy constraints

Innovation

• Privacy-aware DNN model deployment
• Lyapunov-based optimization approach
• Coalition formation game model