基于光学向量矩阵乘法器的光学信息处理系统研究  

An optical digital signal processing system based on optical vector matrix multiplier

作　　者：卢洋洋[1] 周平 朱巍巍[1] 张磊[1] 杨林[1] 

机构地区：[1]中国科学院半导体研究所,集成光电子国家重点实验室,北京100080

出　　处：《光电子．激光》2013年第9期1656-1661,共6页Journal of Optoelectronics·Laser

基　　金：国家自然科学基金(61205008)资助项目

摘　　要：光学向量-矩阵乘法器(OVMM)作为一种利用光学方式进行向量-矩阵运算(VMM)的光学系统,由于采用天然具有高带宽、高并行性的光学处理方式,在海量数据处理领域极具潜力。本文实现了一套基于空间OVMM的光电混合数字信号处理系统,采用自主设计实现的维度为16×16的空间OVMM作为核心运算单元。实验结果显示,系统能够完成76.8G/s乘法累加(MAC)运算,满足实时数据处理对运算速度的需求。系统使用可编程逻辑器件(FPGA)作为电学协处理单元的核心组成部分,因此具有可编程性,可以满足多种不同的应用需求。With the development of information processing fields, such as radar sonar signal processing, image processing, and speech analysis, the quantity of information needed to deal is eye, growing. Because of its bottleneck, such as crossing talk and high power dissipation,traditional electronic technologies can hardly meet the real-time processing of massive data requirements. Optical vector matrix multiplier （OVMM） which is used to per{orm multiplication and accumulation operation is proved to have great potential in the area of massive data processing because of its advantage of high parallelism and large bandwidth. In this paper, we report an optoelectronic hybrid digital signal processing system employing our independently designed optical vector matrix multiplier with the scale of 16 × 16. Experimental results have shown that the system can achieve a calculation rate of 76.8 MAC/s,which fulfills the requirements for processing massive data in real time. The system uses field programmable gate array （FPGA） as the core component of electronic co-processor unit, so that it is reconfigurable, programmable and qualified for kinds of application requirements.

关 键 词：光计算 光学向量-矩阵乘法器（OVMM） 向量-矩阵乘法运算（VMM） 可编程 

分 类 号：TN248.4[电子电信—物理电子学]