| 一体化光电融合计算发展与挑战 |
| |
| 引用本文： | 张天,淡一航,樊泽洋,陈奇,来一航,徐兴元,徐坤.一体化光电融合计算发展与挑战[J].信号处理,2022,38(11):2234-2247. |
| |
| 作者姓名： | 张天 淡一航 樊泽洋 陈奇 来一航 徐兴元 徐坤 |
| |
| 作者单位： | 北京邮电大学信息光子学与光通信国家重点实验室，北京 100876 |
| |
| 基金项目： | 国家自然科学基金资助62171055国家重点研发计划资助2019YFB1803504信息光子学与光通信国家重点实验室（北京邮电大学）基金资助IPOC2020ZT08 |
| |
| 摘 要： | 近年来，军事作战领域在陆、海、空、天、电“五维一体”全方位、跨域协同、信息共享的联合作战样式的实现上付出了巨大的努力。在多功能综合一体化技术中，包括雷达、通信、电子战、导航、敌我识别等多种设备面临着共享射频资源和数据处理资源的问题。海量的数据处理要求计算平台具有超高计算速度和超高能效的特点，而传统的电子计算平台受到“存储墙”、集成电路的时钟频率、带宽、时延等影响，计算速度和能效提升受到很大限制。光子平台具有高速、大带宽、低串扰等特性，再结合电子手段灵活性和易调控的优势，构造光电融合的计算可以有效提升计算速度和能效。为了更好的将光电融合计算应用在多功能综合一体化技术中，本文对光电融合计算的研究现状进行梳理，按照目前的四种典型平台展开讨论，包括片上集成相干平台、片上集成非相干平台、基于光纤系统的光电融合计算、空间光学衍射平台，并在计算大规模化、非线性光学计算实现、光电高效融合三个未来的重要发展趋势进行阐述和展望。
|
| 关 键 词： | 一体化 光电融合计算 光子神经网络 人工智能 |
| 收稿时间： | 2022-07-12 |
Optoelectronic Fused Computing in Multi-functional Integrated Technology: Development and Challenge |
| |
| Affiliation: | State Key Laboratory of Information Photonics and Optical Communications，Beijing University of Posts and Telecommunications，Beijing 100876，China |
| |
| Abstract: | ? ?Recently， in the field of military operations， great efforts have been made to realize the joint operation mode， consisting of omni-directional， cross domain coordination and information sharing with the integration of army， navy， air force， satellite and radio interference. In the multi-function integrated technology， many kinds of equipment， including radar， communication， electronic warfare， navigation， friend-or-foe identification and so on， are faced with the problem of sharing radio frequency resources and data processing resources. Massive data processing requires the computing platform with ultra-high computing speed and energy efficiency. However， the traditional electronic computing platform is limited by the “storage wall”， the clock frequency， bandwidth， delay of integrated circuits， and so on， leading to the greatly limitation of computing speed and energy efficiency. The photonic platform has the excellent characteristics of high speed， large bandwidth， low crosstalk， and so on. By constructing an optoelectronic fused computing platform， which combining the photonic platform and the electronic means with the advantages of flexibility and easy regulation， the computing speed and energy efficiency can be effectively improved. In order to better apply the optoelectronic fused computing platform in the multi-function integrated technology， the research status of the optoelectronic fused computing platform are reviewed. The discussion is carried out according to current four typical platform， including on-chip integrated coherent platform， on-chip integrated noncoherent platform， optoelectronic fused computing platform based on optical fiber system， and spatial optical diffraction platform. Otherwise， three important future development trends， including large-scale computing， the realization of nonlinear computing by photonics， and the efficient fusion of optic and electronic based systems， are also described and prospected. |
| |
| Keywords: | |
|
| 点击此处可从《信号处理》浏览原始摘要信息 |
|
点击此处可从《信号处理》下载全文 |
