Design And Implementation Of An Intelligent IoT Platform Based On DJL

With the rapid development and popularity of 5G Internet technology,Internet of Things applications are mushrooming with the new network features.5G network provides faster data transmission rate,lower network latency and greater connection density for Io T end devices.These huge number of Io T end devices have higher requirements for the design and implementation of Io T platforms.On the one hand,Io T devices are limited by their size,application environment,power consumption and other factors,which cannot provide themselves with the huge arithmetic power required for artificial intelligence;on the other hand,Io T devices also have a huge amount of device information and data that can provide the data base for intelligent computing.This paper first analyzes the functional requirements and performance requirements of intelligent Io T platform,summarizes and introduces the design cases of existing Io T platforms,analyzes their design architectures,combines the general characteristics of Io T platforms and Io T terminal devices,and specifically analyzes the functional requirements and performance requirements needed for intelligent Io T platforms.Then a new platform architecture design scheme is proposed to design the data acquisition and transmission scheme with reference to various data transmission protocols;the Deep Java Library and distributed microservice architecture are used to realize the splitting and development of functional services of each part of the platform.Then we improve and implement video compression algorithm,based on H.264 video compression algorithm and blind super-resolution reconstruction algorithm based on unsupervised learning,to improve video transmission efficiency while ensuring video quality through quality reduction transmission and super-resolution reconstruction.Finally,the distributed cluster deployment method is used to complete the deployment of TCP connection end,gateway,platform,video push stream,deep learning module and other functions,and access to Io T terminal devices to verify the practical application value of this paper for Io T platform design and implementation research.