Beijing Unicom GSM Mobile Telephone Traffic Prediction Based On Neual Network

Starting with the basic theory of neural network, this thesis studies the GSM traffic forecasting based on neural network and study the various factors on the impact of traffic. The major work done are as follows.(1) The long-term traffic prediction for GSM based on artificial neural network:This thesis analyzes the factors of influencing traffic and sets up prediction model of the traffic of GSM with BP neural network. In this thesis we make multiple performance analytical comparison among a few of kinds of function and some BP network structures by experiment and finally determine network model, which shows a satisfying long-term traffic prediction result.(2) Study the various factors on the impact of GSM traffic:This thesis uses recycling methods to identify the various factors on the impact of traffic, to analysis the importance of various factors qualitatively and to identify which are the key factors, and remove the unsuitable input values for a BP neural network simulation. Then we simulate the model with MATLAB without the unsuitable input values in order to define the network structure and the training function again to adapt to the network model without the unnecessary factors.(3) The short-term traffic prediction for GSM based on Elmen artificial neural network:This thesis studies the characteristics of algorithm of the Elman neural network to select the appropriate input and output data and achieve a better short-term forecasting results through training and simulation.Accurate predictions can prevente major issues such as traffic congestion, low through rate caused by excessive traffic. In terms of long-term or short-term, forecasting network traffic has an important significance for the healthy and stable development of the GSM communication network and reducing the risk of systems.