Machine Intelligence on Wireless Edge Networks

2025/06/13 by Sri Krishna Vadlamani, Kfir Sulimany, Vadlamani, Sri Krishna +7
Computer Science · Engineering · #Emerging Technologies (cs.ET) #Energy Efficient Wireless Sensor Networks #FOS: Computer and information sciences #Machine Learning (cs.LG) #Opportunistic and Delay-Tolerant Networks #Wireless Body Area Networks

paper · pdf · doi:10.48550/arxiv.2506.12210

openalex publication_date 2025/06/13 · openalex created_date 2025/10/13 · openalex updated_date 2026/07/28

Abstract

Machine intelligence on edge devices enables low-latency processing and improved privacy, but is often limited by the energy and delay of moving and converting data. Current systems frequently avoid local model storage by sending queries to a server, incurring uplink cost, network latency, and privacy risk. We present the opposite approach: broadcasting model weights to clients that perform inference locally using in-physics computation inside the radio receive chain. A base station transmits weights as radio frequency (RF) waveforms; the client encodes activations onto the waveform and computes the result using existing mixer and filter stages, RF components already present in billions of edge devices such as cellphones, eliminating repeated signal conversions and extra hardware. Analysis shows that thermal noise and nonlinearity create an optimal energy window for accurate analog inner products. Hardware-tailored training through a differentiable RF chain preserves accuracy within this regime. Circuit-informed simulations, consistent with a companion experiment, demonstrate reduced memory and conversion overhead while maintaining high accuracy in realistic wireless edge scenarios.

Citations

• Disaggregated Deep Learning via In-Physics Computing at Radio Frequency
• Transformers without Normalization
• QAMNet: Fast and Efficient Optical QAM Neural Networks
• Quantum-secure multiparty deep learning
• Training of Physical Neural Networks
• Arbitrary control over multimode wave propagation for machine learning
• Wireless Distributed Matrix-Vector Multiplication using Over-the-Air Computation and Analog Coding
• AirNN: Over-the-Air Computation for Neural Networks via Reconfigurable Intelligent Surfaces
• The physics of optical computing
• Nonlinear Processing with Linear Optics
• Powering AI at the Edge: A Robust, Memristor-based Binarized Neural Network with Near-Memory Computing and Miniaturized Solar Cell
• Backpropagation-free Training of Deep Physical Neural Networks
• Computing Functions Over-the-Air Using Digital Modulations
• A Survey on Over-the-Air Computation
• RF-Photonic Deep Learning Processor with Shannon-Limited Data Movement
• MatPIM: Accelerating Matrix Operations with Memristive Stateful Logic
• Single-Shot Optical Neural Network
• All-Photonic Artificial Neural Network Processor Via Non-linear Optics
• MultPIM: Fast Stateful Multiplication for Processing-in-Memory
• Programmable photonic circuits
• Language Models are Few-Shot Learners
• Memory devices and applications for in-memory computing
• All-optical spiking neurosynaptic networks with self-learning capabilities
• PUMA: A Programmable Ultra-efficient Memristor-based Accelerator for Machine Learning Inference
• Split learning for health: Distributed deep learning without sharing raw patient data
• All-Optical Machine Learning Using Diffractive Deep Neural Networks
• Efficient Processing of Deep Neural Networks: A Tutorial and Survey
• Efficient Processing of Deep Neural Networks: A Tutorial and Survey
• Neuromorphic Silicon Photonic Networks
• Quantized Neural Networks: Training Neural Networks with Low Precision Weights and Activations
• Layer Normalization
• Robust Analog Function Computation via Wireless Multiple-Access Channels
• The Emergence of Edge Computing

Related

• Disaggregated Deep Learning via In-Physics Computing at Radio Frequency
• Communication-Efficient Edge AI Inference Over Wireless Networks
• Wireless Network Intelligence at the Edge
• An Overview of Data-Importance Aware Radio Resource Management for Edge Machine Learning
• Edge Intelligence Over the Air: Two Faces of Interference in Federated Learning