Research Of Cellular Mobile Network Performance Analyzing Platform And Performance Detecting

With the rapid development of cellular wireless network and the exponential growth of traffic data,the intelligent terminals are widely used,users' demands for the quality of service are increasing gradually.Therefore,telecommunication operators and hardware manufacturers should not only ensure the quality of cellular network but also optimize the network performance continuously.Among all the solutions,the evaluation and prediction of the performance of cellular mobile work plays an important role in the optimization of network performance and the overall planning of telecommunication network.In addition,the android system has become the operating system of most intelligent devices with high market share and good representativeness.In this paper,we built a platform for cellular network data collecting and analyzing based on the android operating system,and accurately located the cellular network performance data.We also proposed the network performance analyzing and predicting strategies.The main contributions of this paper are summarized as follows:1.We built a cellular network data collecting and analyzing platform NetAnalyzer,and introduced the overall system frame and the main components.The client system of our platform collects multiple dimensions of cellular network,for example signal to interference and noise ratio,transport delay,signal strength and so on.The server system of our platform takes charge of data multithreading transporting,data caching and calculating.Our platform also provides data visualization function,from which we can observe the location distribution of cellular network data clearly.Besides,A hybrid indoor location algorithm is proposed,and is utilized in our platform.We deployed the Bluetooth devices and sensors indoor,and use our platform to collect and locate the client terminal data.By utilizing our location algorithm,we can get the position distribution of indoor cellular network performance parameters with high accuracy.2.We proposed a cellular network performance analyzing and evaluating strategy and utilize the strategy on the performance analyzing module of our platform.Based on the multiple dimensions data of network performance collected by the platform like signal strength,data rate and so on,we evaluated the cellular network performance by the iterating and weighting the data spatially first,and use the contour detection algorithm to achieve the interference elimination of the result.Besides,we proposed an integrated indicator to evaluate the network performance.We divided all the network performance indicators into different grades.By utilizing the network performance evaluating strategy,we can detect and observe the distribution of the network performance,and achieve a higher precision by using the interference elimination method.3.We built a time series prediction model and applied it to our platform to predict the temporal performance sequence of cellular network.The proposed model is composed of the DBN(Deep Belief Network)model and the improved SVR(Support Vector Regression)algorithm,the DBN model is utilized to extract the feature of the time series and the improved SVR algorithm is utilized to predict the obtained feature sequences.In this paper,we mainly predict the signal strength sequence.By comparing the predicted results with the actual sequence,we can found that the predicted results of proposed prediction model can achieve high accuracy.Besides,by improving the SVR algorithm,our proposed predicting model runs faster,and the accuracy of our proposed model can achieve 93.64%by optimizing the number of data input dimension.