Base Station Network Traffic Prediction Method And System Implementation For Smart City

With the rapid development of science and technology and the continuous improvement of people's quality of life,smart cities supported by cloud computing,Internet of things and mobile Internet are expected to provide people with intelligent and modern living environment.In the smart city,the demand and stickiness of users for mobile Internet are increasing,and the service quality of network is facing a huge challenge.As an important carrier of network traffic,the base station has grasped the trend of base station traffic trend,and allocated network resources such as network upgrading and optimization,network traffic unloading,etc.in a timely manner,which plays an important role in improving the quality of network service.Therefore,in the context of smart city,this paper designs a base station network traffic prediction method and system for smart city.The specific research contents are as follows:First of all,combined with the current research on tactile Internet and the vision of smart city development,a smart city system framework based on tactile Internet is proposed.Tactile Internet is a kind of communication infrastructure with ultra-reliable and low latency communications.The introduction of tactile Internet into smart city not only innovates the existing communication and interaction methods,but also is expected to make use of tactile information and tactile related applications,so as to better promote the intelligent process of smart city and the experience quality of end users.Then,according to the traffic data of the base station in the smart city,the data preprocessing work such as outliers processing,missing values processing and data standardization is carried out;the traffic characteristics of the base station are extracted through the analysis of the time characteristics of the traffic data of the base station;after the feature engineering is completed,the Seasonal-Trend decomposition procedure based on Loess is used to integrate with the Long ShortTerm Memory(STL-LSTM).The original base station traffic data is decomposed to remove the randomness component,and its long-term and short-term dependence characteristics are further combined with highly predictable regularity and unpredictable randomness in traffic data.The experimental results show that STL-LSTM has better traffic prediction ability than LSTM.Finally,in order to ensure that the operation and maintenance personnel can get the change trend of the base station traffic in time and make corresponding adjustments to the base station load situation,a base station traffic prediction system based on spark big data processing platform and gentellella front-end display framework is designed,which realizes the functions of system login,system management,cluster management and maintenance,traffic prediction and system fault feedback.