Goal-oriented Communications based on Recursive Early Exit Neural Networks

2024-12-27

AI Summary

To address the fundamental trade-off among latency, inference accuracy, and resource consumption in goal-oriented semantic communication, this paper proposes a dynamic computation offloading framework based on recursive early-exit neural networks. Our method introduces two key innovations: (1) a novel inter-layer confidence recursion mechanism enabling fine-grained, adaptive early-exit decisions during inference; and (2) a reinforcement learning–based online optimizer that jointly considers wireless channel state, inference accuracy, and resource cost to coordinate edge–cloud inference task scheduling. Experimental evaluation in edge intelligence scenarios demonstrates significant improvements in the three-way trade-off: average end-to-end latency is reduced by 37%, device energy consumption decreases by 42%, while maintaining an inference accuracy of ≥98.5%. These results consistently outperform state-of-the-art baseline approaches.

Technology Category

• Machine Learning: Learning on the Edge & Model Compression
• Search and Optimization: Learning to Search
• Planning, Routing, and Scheduling: Planning with Language Models

Application Category

• Economics, Online Markets and Human Computation: Incentives in network design for Web infrastructures and ecosystems
• Systems and Infrastructure for Web, Mobile and WoT: Energy management for devices in mobile Web and WoT environments
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation

Abstract

This paper presents a novel framework for goal-oriented semantic communications leveraging recursive early exit models. The proposed approach is built on two key components. First, we introduce an innovative early exit strategy that dynamically partitions computations, enabling samples to be offloaded to a server based on layer-wise recursive prediction dynamics that detect samples for which the confidence is not increasing fast enough over layers. Second, we develop a Reinforcement Learning-based online optimization framework that jointly determines early exit points, computation splitting, and offloading strategies, while accounting for wireless conditions, inference accuracy, and resource costs. Numerical evaluations in an edge inference scenario demonstrate the method's adaptability and effectiveness in striking an excellent trade-off between performance, latency, and resource efficiency.

Problem

• Semantic Communication
• Efficiency Improvement
• Resource Optimization

Innovation

• Dynamic Task Allocation
• Reinforcement Learning Optimization
• Edge Computing Efficiency