C-RAN Radio Resource Management Based On Big Data

At present,wireless communication technology is developing at a high speed,the application of wireless communication is more and more frequent,the coverage is wider and wider,the requirements for communication quality are getting higher and higher,and the number of wireless access devices is increasing.Along with these tremendous changes,the whole society expects that wireless communications will have a broader distribution,lower price and better service in the future.C-RAN is proposed to meet needs in the field of wireless communications.The centralized management of C-RAN brings more possibilities to the network.Due to centralized management,a large amount of wireless information is generated,and massive data brings new opportunities.The main context and work are as follows.First,the SINR prediction based on C-RAN are studied.Based on the C-RAN network architecture,this thesis realizes the prediction of uplink SINR by the wireless data in the network.The effectiveness of using the neural network algorithm to predict the uplink SINR of the user through user occupancy are verified.Secondly,a scheduling algorithm for accelerating the convergence of SINR prediction is proposed.This thesis proposes a method for detecting the accuracy of interference prediction between users.This method uses the convergence fluctuation of uplink SINR predicted result in the single-user interference scenario of the neural network to screen out the interference users whose interference prediction is inaccurate.After identifying the user with inaccurate interference prediction,the user scheduling policy aiming at improving the convergence speed of the uplink SINR prediction of the user is used for user scheduling and the user who is inaccurately predicted is preferentially scheduled.This scheduling algorithm effectively reducing the sample data armount requirement of the training set and accelerates convergence of uplink SINR predictions.The simulation proves that this algorithm can effectively improve the convergence speed of uplink SINR prediction.Third,a simulation platform for verifying the above algorithm was built.The simulation platform built in this thesis is a system-level dynamic simulation platform,which is dynamic and can simulate real scenes in the real wireless network.The simulation platform built in this thesis can simulate three basic network nodes:user,home base station and macro base station.In the data transmission process,the SINR is calculated according to the co-channel interference,fast fading and path loss.The simulation platform also supports HARQ function.And integrated the conventional maximum carrier-to-interference ratio scheduling algorithm,Round-Robin scheduling algorithm and proportional fair scheduling algorithm.Under the premise of LTE simulation function,this simulation platform adds C-RAN related modules,mainly for data storage and machine learning.