Research On Resource Allocation Optimization Of Wireless Communication Network Based On User Behavior Prediction

To meet the needs of users,today's mobile network deployment is based on the maximum demand of users in the region.In one day,the demand for mobile resources in a cell fluctuates with time.When the user's need is low,it often results in a large waste of resources.In future networks,with the increase of mobile devices,this waste phenomenon will be more obvious if the base station provides all the resources for the cell according to the maximum demand every moment of the time.Big data and machine learning have been applied to all aspects of life.In this paper,we consider that the application of machine learning algorithm to the research of mobile user behavior data,the information excavated from the user behavior information is very valuable to optimize the allocation of mobile communication network resources.The main contributions of this thesis are showed in following:(1)The machine learning algorithm is applied to the data research in the field of communication.In this thesis,the real-time data of the number of users in the cell is obtained.Through the parameter adjusted random forest,support vector machine and the long short-term memory network constructed in the thesis,the thesis predicts the number of users in the next time in the cell and prove the superiority of the proposed algorithm by comparing the prediction results.(2)The analysis of user behavior is applied to the resource allocation of actual single-cell communication scenarios to optimize energy efficiency.In a single-cell large-scale MIMO(Multiple-Input Multiple-Output)scenario,according to the load prediction result of the previous step,adjusting the number of antennas turned on by the base station at the next moment to achieve the maximum energy efficiency.(3)Self-Organizing Network Multi-cell Collaborative Optimization.The mobile edge computing server is used to collect user information data in the cell to implement the prediction algorithm and upload the information to the core network.In the multi-cell scenario,this thesis proposes a self-optimizing scheme for multi-cell self-organizing networks.According to the predicted future load in each cell,the number of antennas of each cell is cooperatively optimized.Through the dynamic optimization scheme of the number of antennas in each cell,energy efficiency can be improved by about 120%and energy saving by about 60%compared with that when all the antennas are fully open.Since this article uses a real data set,it is of certain reference significance for operators to optimize the allocation of radio resources in multiple cells.