Disaggregated Deep Learning via In-Physics Computing at Radio Frequency

2025-04-24

AI Summary

Real-time deep learning inference on resource-constrained edge devices faces severe energy consumption and memory bottlenecks. To address this, we propose a wireless broadcast–driven, RF-domain native complex matrix–vector multiplication architecture—the first to jointly enable over-the-air wireless distribution of model weights and direct execution of complex linear operations at the RF physical layer, thereby decoupling computation and storage across devices. Implemented on a software-defined radio platform, our design integrates over-the-air weight broadcasting, analog-domain complex multiply-accumulate (MAC) operations, and dedicated signal processing circuitry. Evaluated on image classification, it achieves 95.7% accuracy with only 6.0 fJ/MAC per client and an energy efficiency of 165.8 TOPS/W—exceeding conventional digital implementations by over two orders of magnitude—thus significantly advancing the energy-efficiency frontier for edge AI.

Technology Category

• Machine Learning: Learning on the Edge & Model Compression
• Application Domains: Internet of Things, Sensor Networks & Smart Cities
• Planning, Routing, and Scheduling: Optimization of Spatio-temporal Systems

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Energy management for devices in mobile Web and WoT environments
• Search and Retrieval-Augmented AI: Efficiency and scalability of Web search engines
• Security and Privacy: Security and privacy of machine learning and AI applications

Abstract

Modern edge devices, such as cameras, drones, and Internet-of-Things nodes, rely on deep learning to enable a wide range of intelligent applications, including object recognition, environment perception, and autonomous navigation. However, deploying deep learning models directly on the often resource-constrained edge devices demands significant memory footprints and computational power for real-time inference using traditional digital computing architectures. In this paper, we present WISE, a novel computing architecture for wireless edge networks designed to overcome energy constraints in deep learning inference. WISE achieves this goal through two key innovations: disaggregated model access via wireless broadcasting and in-physics computation of general complex-valued matrix-vector multiplications directly at radio frequency. Using a software-defined radio platform with wirelessly broadcast model weights over the air, we demonstrate that WISE achieves 95.7% image classification accuracy with ultra-low operation power of 6.0 fJ/MAC per client, corresponding to a computation efficiency of 165.8 TOPS/W. This approach enables energy-efficient deep learning inference on wirelessly connected edge devices, achieving more than two orders of magnitude improvement in efficiency compared to traditional digital computing.

Problem

• Reducing energy use in edge deep learning
• Enabling efficient wireless model broadcasting
• Performing RF in-physics matrix computations

Innovation

• Disaggregated model access via wireless broadcasting
• In-physics computation at radio frequency
• Ultra-low power deep learning inference