| The IOV(Internet of vehicles)business not only includes traditional telematics services such as security,remote control and voice navigation,but also includes innovative services which is based on big data analysis such as fleet management,driving behavior analysis and vehicle appraisal and evaluation.The collection and distribution of vehicle data is the cornerstone of all these businesses.With the increasing number of vehicles,and the types of vehicle data to be collected are constantly increasing,so the number of vehicle data increases with the number of billions.Problems need to be solved such as the performance of vehicle data collection and storage,and managing the tasks of collection and distribution according to business service types.In view of the above problems,based on the relevant technologies of microservices architecture,micro-service monitoring and Kafka middleware,this paper designs and implements a micro-service-based vehicle data collection and distribution system(referred to as VDCD system),which is on the background of the actual project of a IOV service company which occupies the leading position in China.Key features of the system include Kafka-based vehicle data collection and distribution,Cassandra-based vehicle data storage,and Spring Cloud Sleuth-based service monitoring.Test results show that the system is feasible and effective.The main points of this paper are as follows:1)Kafka-based vehicle data collection and distribution.Vehicle data collection needs to support thousands of data types and multiple transport protocols such as SOAP,REST,JMS,etc.The collected vehicle data needs to be distributed to multiple downstream business systems according to the service type.This paper presents a micro-service-based vehicle data collection and distribution task management program,based on Spring Cloud and Kafka framework.The test shows that the VDCD system can support the collection and distribution of 100,000 vehicle data per second of nearly 50 megabytes.2)Cassandra based vehicle data storage.Vehicle data storage needs to support the uploading of data which is generated from millions of active vehicles in the rush hour,to support data query according to vehicles,time and type of business.This paper designs and implements a distributed vehicle data storage module based on Cassandra.Test results show that VDCD system can support 100,000 hybrid read and write scenes per second,writing and reading delay of 99.9% are within 300 milliseconds.3)Service Monitoring based on Spring Cloud Sleuth.Services for vehicle data collection,distribution and storage need to be highly available and monitor the lifecycle of service requests.This paper designs and implements a subsystem to realize the service monitoring,based on open-source framework such as Spring Cloud Sleuth and Zipkin.The test results show that when the sampling rate of mission service monitoring is set to 0.1,the throughput of mission service is reduced by about 0.5% and the response delay is increased by about 1%.The VDCDS system can be used in any IOV business that requires the vehicle data.Such as historical track service based on vehicle position data;driving behavior analysis service based on vehicle engine on/off and speed data;and vehicle appraisal and evaluation service based on vehicle diagnostic and collision data.The design and construction of VDCDS system can enhance the cohesion of each system of IOV platform while keeping each system loosely coupled.So as to reduce the cost of each system in carrying out new business functions and maintenance,also to enhance the robustness and availability of the whole IOV platform. |
