摘要
鉴于可见光屏幕通信具有抗干扰能力强、不占用频谱资源和链路部署简单易于交互等特点,设计了基于卷积神经网络(CNN)的可见光屏幕通信系统。重点阐述了帧结构的定义、接收单元CNN模块的引入以及解析机制的设计。帧结构的定义确保了整个系统的可靠性,丰富了屏幕通信传输内容的多样性;CNN模块的引入使接收单元可以自动识别屏幕发送的内容,不依赖传统的定位检测图形,提高了智能化和信息携带量;两种解析机制的设计提高了屏幕通信的普适性。实验系统实现了96.4%的识别成功率,达到实时150kbit/s和非实时300kbit/s的通信速率,可以传输文本、图片和音频等类型的文件。
Visible light screen communication has the characteristics of strong anti interference ability, no occupation of frequen- cy spectrum resources, simple link deployment, and easy interaction. This paper designs a visible light screen communication system based on Convolutional Neural Network （CNN）. It includes the definition of the frame structure, the introduction of the receiving unit CNN module, and the design of the resolution mechanism. The definition of the frame structure ensures the reliability of the system and enriches the diversity of the transmission content of the screen communication. The introduction of the CNN module allows the receiving unit to automatically identify the content sent by the screen. It also does not rely on traditional positioning detection patterns, which improves the intelligence and information carrying capacity. The design of the two resolution mechanisms improves the universality of screen communication. It can achieve 96.4% recognition success rate at real time 150 kbit/s and non real time 300 kbit/s communication rate, which can be used to transmit text, picture, audio type files.
作者
刘文楷
徐一鸣
武梦龙
LIU Wen-kai;XU Yi-ming;WU Meng-long(College of Computing;College of Electronic Information Engineering,North China University of Technology,Beijing 100144,China)
出处
《光通信研究》
北大核心
2018年第6期42-46,共5页
Study on Optical Communications
基金
国家自然科学基金资助项目(61471052)
关键词
可见光通信
屏幕通信系统
卷积神经网络
解析机制
visible light communication
screen communication system
CNN
resolution mechanism
