| With the development of the Internet of Things technology,the massive data generated by intelligent devices has challenged the original cloud computing service model,edge computing emerged in this historic moment.Gateway is an important part of the Internet of Things system,as the deployment location is closer to the edge of the network,it's communication and data processing capabilities need to be improved.Based on microservices related technologies,this paper designed and constructed a gateway system for edge computing,the main work is as follows:Firstly,in response to the interaction between the edge computing gateway with the northbound cloud platform and the southbound device,an edge computing gateway based on microservices is designed,the gateway is mainly composed of basic service components,communication management components and device management components.The basic service components provide the underlying support for the gateway.The communication management component provides communication services between the gateway and the cloud,the gateway and the device,and the internal components of the gateway.The device management component provides device access and management services.Secondly,according to the business characteristics of the edge computing gateway,a sliding-window-based response ratio polling algorithm is proposed,and the Java native thread pool is optimized in the device management microservice.The former solves the problem of determining the initial weight of the original algorithm by introducing the response ratio parameter,the latter improves the processing capacity of the thread pool in high-concurrency scenarios by refactoring the work queue and workflow of the Java native thread pool.Finally,build the edge computing gateway based on the system's overall architecture.The gateway service is deployed and run as a Docker image through an automated deployment platform.The function of the gateway and the performance of key components are tested to verify the feasibility of business functions and the effectiveness of strategies such as improvement of load balancing algorithms,optimization of thread pools,and introduction of object pools. |
