| With the development of radio technology and the rise of mobile Internet,a large number of portable and removable monitoring terminal has gradually become the main force of spectrum monitoring.The expansion of network has brought a series of data transmission,data storage and data analysis and processing issues.It is in this background,from the perspective of spectral data,this thesis studies the issues as follows: data transmission under network congestion,mass spectrum data file storage,and a large number of spectrum data processing tasks scheduling.It provides a guideline for future spectrum monitoring.Based on the general requirements of the spectrum monitoring system,this paper proposes a distributed service cluster architecture to face the the large data amount and high concurrency scenes.According to the different roles,the server is divided into monitoring server,file server,database server,computing server,web server.Each server can provide services in the form of cluster,which effectively enhance the service ability and extensibility of the system.Secondly,on the basis of this system,this paper proposes a congestion control mechanism based on data transmission delay,data priority,the type and working state of monitoring terminal.The mechanism can automatically extract the files to control the data transmission rate according to the degree of congestion,then it can alleviate the network congestion while maintaining the effect of spectrum monitoring.For the mass spectrum data,this paper implements the distributed storage with Redis and MongoDB.Use Redis to build a distributed cache system to solve the problem of real-time data writing by the monitoring terminal with high concurrency,and store the latest data in memory to speed up the reading and writing.Use the MongoDB database that supports automatic sharding and flexible expansion to implement persistent storage of spectrum data.Finally,in view of the current system spends a lot of time on the remote data reading,this paper proposes a dynamic delay scheduling algorithm based on task classification,which uses fuzzy comprehensive evaluation method to divide the task into three categories.Each category of tasks corresponding to the waiting time threshold is different,so that the task in the waiting time threshold can be scheduled to the computing node which contains the task required data as much as possible,so as to avoid the time cost on the remote data reading. |
