Research On Real-Time Data Transmission And Distribution For Embedded Systems

Real-time data transmission and distribution is one of the key technologies of embedded systems.Distributing real-time data to corresponding processing platforms at high speed in various distributed embedded systems can effectively use the processor's computing ability,which is of great value and significance for applications such as video surveillance and signal processing,etc.This thesis first analized the embedded communication protocol RapidIO and 10G Ethernet protocols in depth.Based on this,it proposed data transmission and distribution software-hardware design solutions for three embedded platforms,and implemented high-performance real-time data transmission and distribution in their respective application scenarios.The main tasks completed in this thesis are as follows:1)A RapidIO real-time communication network was built for the RapidIO exchange platform and then implemented RapidIO enumeration function through FPGA,so that the processor chips in and between the boards can transmit and distribute real-time data through the RapidIO transmission link.2)For the x86 embedded platform,a "PCIe+10 Gigabit Ethernet" high-speed link real-time data forwarding scheme was proposed,and then a multi-threaded data transmission control application was designed to realize real-time signal data transmission and transmission between multiple processing platforms.3)A 10-Gigabit Ethernet data communication link was designed for the real-time storage platform,then file transfer server and client program based on the UDP protocol were developed.Finally a high-speed data transmission and distribution function was realized for the real-time storage platform.Experimental solutions were designed respectively for the above three platforms to test high-performance data transmission and distribution.The results showed that the RapidIO transmission link latency based on RapidIO ID data distribution and the "PCIe+10 Gigabit Ethernet" transmission speed,UDP download speed based on Ethernet IP address data distribution meet the expected requirements.