Design And Implementation Of Software Defined Network Traffic Situational Awareness System Based On ONOS

Network traffic situation awareness can quickly and accurately reflect the current running status of the network traffic,and can also predict the future trend of the network traffic operation.It is an important means of network traffic management.Combined with the characteristics of SDN network and ONOS controller,this paper designs and implements a network traffic situation awareness system,which can be divided into traffic situation index extraction subsystem,traffic situation assessment subsystem and traffic situation prediction subsystem.The main work of this paper is as follows:Firstly,this paper designs and completes the traffic situation index extraction subsystem.Considering the completeness and accessibility,the traffic situation index system is constructed hierarchically.And then,the traffic situation index collection module is designed and implemented to access the ONOS controller every certain time to obtain the original data and process the traffic situation index.Lastly,the data storage module is designed and implemented with MySQL,which can easily add,delete and query the data.Secondly,the traffic situation assessment subsystem is designed and implemented.This paper proposes a method of evaluation grading design,defines the practical meaning of each evaluation grade in advance,which can reduce the network managers' learning cost.And this paper designs and implements a traffic situation assessment algorithm based on supervised machine learning methods BP and RBF.Thirdly,the traffic situation prediction subsystem is designed and implemented.This paper proposes an improved situation prediction process,which predicts the traffic situation index,and then evaluates the situation of the obtained index prediction value to obtain the traffic situation evaluation prediction value.Among them,considering the high independence of the indicators,the method of "separation prediction re combination" is proposed to predict the situation indicators.In addition,the traffic situation index prediction algorithm based on DNN,GRU,SVR,GWO-SVR and PSO-SVR is designed and implemented.Finally,the experimental platform is built based on Mininet and ONOS controller,and the data traffic service is designed and implemented.Based on the above-mentioned platforms and services,the traffic situation assessment algorithms and the traffic situation indicator prediction algorithms designed in this paper are compared and tested.Considering accuracy and training time,this paper selects the RBF-based traffic situation assessment model and the PSO-SVR-based traffic situation index prediction algorithm.And then the network traffic situation assessment and prediction functions are verified through experimental tests.