Research And Application Of Operation Quality Comprehensive Assessment Method For Video Conferencing System

Video conferencing system provides users in different places with a remote communication and collaboration mode that is not restricted by space and geographical factors.This system not only improves the coordination efficiency in enterprise,but also avoids the travel and conference costs brought by organizing on-site meetings.So it has become an important supporting platform for the efficient operation of enterprises in different places.With the continuous promotion of enterprise information construction and high-definition video conference,the video conferencing system's safety and stability are facing severe challenge.Therefore,it is urgent to establish a comprehensive assessment model for periodical quantization of video conferencing system,which can help managers understand system's operation quality in each dimension,so as to locate system's current weaknesses and develop improvement plans.At present,the operation quality assessment models for video conferencing system usually take serveral evaluation indicators at network and business level as input,and consider the user's subjective satisfaction as expected output.However,these assessment models not only ignore the daily operation and maintenance's protection on system's stability,but also lack the mechanism of extracting knowledge related to system optimization.Therefore,this paper focuses on video conferencing system's influence factors in network,business,and operation and maintenance aspects,then designs and implements the operation quality assessment method for video conferencing system.The specific research contents are as follows:1)We propose the operation quality assessment method based on genetic regularization BP neural network.This method consists of four stages:construction of evaluation indicator system,feature selection,calculation of system operation quality level and approximate knowledge extraction.Firstly,we build a system,operation quality's evaluation indicator system for video conferencing system based on its operational statistics,which covers network,business,and operation and maintenance aspects.Then,we perform feature selection on the candidate feature set composed of all three-level evaluation indicators,and use the mutual information as measure to select the optimal indicator subset,which is entered into assessment model to analyze video conferencing system's operation quality level.Finally,we use Casual Index method to extract approximate knowledge from the trained assessment model,which assists mangers in developing system optimization plan.In order to make up for traditional BP neural network in overfitting and trend of falling into local optima,we introduce L2 regularization term into neural network's cost function,and use genetic algorithm to search initial parameter closer to global optima for neural network.Simulation experiments show that the assessment model in this paper has advantages in convergence speed,prediction accuracy and generalization ability.2)We design and develop the video conference operation quality assessment system,which provides support for all stages of comprehensive assessment method in this paper through three core modules:evaluation indicator system management,assessment model management and operation quality assessment.Firstly,managers can perform evaluation indicator management and feature selection to construct the evaluation indicator system,which is taken as assessment model's input.Then,managers can provide the legal basic data and algorithm parameters to perform multiple trainings on the assessment model,and query the approximate knowledge extracted from the latest model after each training process completes.When the model's classification performance on training set and test set meet managers' requirements,managers and operation personnels can choose one day or several days as time dimension of operation quality assessment,then query and export video conferencing system's operation quality assessment report for one day,or the system operation quality's change tendency report for a period of time.In the experiment using real data from video conferencing system in a provincial power company,each functional module of this assessment system can operate correctly and steadily.