2020/09/03 by Nicolas Skatchkovsky, Skatchkovsky, Nicolas, Hyeryung Jang +3 · 1 citation
Computer Science · Engineering · #Advanced Memory and Neural Computing #FOS: Computer and information sciences #FOS: Electrical engineering #Information Theory (cs.IT) #Machine Learning (cs.LG) #Neural Networks and Reservoir Computing #Neural and Evolutionary Computing (cs.NE) #Signal Processing (eess.SP) #Underwater Vehicles and Communication Systems #electronic engineering #information engineering
paper · pdf · doi:10.48550/arxiv.2009.01527
openalex publication_date 2020/09/03 · openalex created_date 2022/07/25 · openalex updated_date 2026/07/28
This paper introduces a novel "all-spike" low-power solution for remote\nwireless inference that is based on neuromorphic sensing, Impulse Radio (IR),\nand Spiking Neural Networks (SNNs). In the proposed system, event-driven\nneuromorphic sensors produce asynchronous time-encoded data streams that are\nencoded by an SNN, whose output spiking signals are pulse modulated via IR and\ntransmitted over general frequence-selective channels; while the receiver's\ninputs are obtained via hard detection of the received signals and fed to an\nSNN for classification. We introduce an end-to-end training procedure that\ntreats the cascade of encoder, channel, and decoder as a probabilistic\nSNN-based autoencoder that implements Joint Source-Channel Coding (JSCC). The\nproposed system, termed NeuroJSCC, is compared to conventional synchronous\nframe-based and uncoded transmissions in terms of latency and accuracy. The\nexperiments confirm that the proposed end-to-end neuromorphic edge architecture\nprovides a promising framework for efficient and low-latency remote sensing,\ncommunication, and inference.\n