{"title":"Integrated Photonic Convolution Accelerator Empowered by Thin-Film Lithium Niobate Modulators","authors":"Haojun Zhou;Bo Wu;Shiji Zhang;Mengyue Xu;Jingyi Wang;Hailong Zhou;Xinlun Cai;Jianji Dong","doi":"10.1109/LPT.2025.3541048","DOIUrl":null,"url":null,"abstract":"The current optical convolution architectures are facing challenges related to limited scalability, excessive data redundancy and restricted processing bandwidth. In this work, we introduce an integrated photonic convolution accelerator (IPCA) empowered by high-speed thin-film lithium niobate (LN) modulators. Consequently, data replication redundancy is free and Fourier transform is avoided, which paves the way for highly efficient convolution scaling. We implement convolution in an optical frequency spacing of 8 GHz with a power consumption of only 52.9 mW. Optical neural networks of different parameter sizes are constructed across various complex tasks. Given its advantages to address energy consumption and computing power challenges inherent to current AI advancements, our method heralds a pivotal shift in upcoming optical computing hardware architectures.","PeriodicalId":13065,"journal":{"name":"IEEE Photonics Technology Letters","volume":"37 7","pages":"385-388"},"PeriodicalIF":2.3000,"publicationDate":"2025-02-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"IEEE Photonics Technology Letters","FirstCategoryId":"5","ListUrlMain":"https://ieeexplore.ieee.org/document/10879569/","RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q2","JCRName":"ENGINEERING, ELECTRICAL & ELECTRONIC","Score":null,"Total":0}
引用次数: 0
Abstract
The current optical convolution architectures are facing challenges related to limited scalability, excessive data redundancy and restricted processing bandwidth. In this work, we introduce an integrated photonic convolution accelerator (IPCA) empowered by high-speed thin-film lithium niobate (LN) modulators. Consequently, data replication redundancy is free and Fourier transform is avoided, which paves the way for highly efficient convolution scaling. We implement convolution in an optical frequency spacing of 8 GHz with a power consumption of only 52.9 mW. Optical neural networks of different parameter sizes are constructed across various complex tasks. Given its advantages to address energy consumption and computing power challenges inherent to current AI advancements, our method heralds a pivotal shift in upcoming optical computing hardware architectures.
期刊介绍:
IEEE Photonics Technology Letters addresses all aspects of the IEEE Photonics Society Constitutional Field of Interest with emphasis on photonic/lightwave components and applications, laser physics and systems and laser/electro-optics technology. Examples of subject areas for the above areas of concentration are integrated optic and optoelectronic devices, high-power laser arrays (e.g. diode, CO2), free electron lasers, solid, state lasers, laser materials'' interactions and femtosecond laser techniques. The letters journal publishes engineering, applied physics and physics oriented papers. Emphasis is on rapid publication of timely manuscripts. A goal is to provide a focal point of quality engineering-oriented papers in the electro-optics field not found in other rapid-publication journals.