Research On Key Technologies Of AMC For LTE Uplink

Adaptive modulation and coding(AMC)technology is widely used in LTE wireless communication systems to cope with the time-varying characteristics of wireless channels.Based on the channel state information(CSI),AMC technology selects the optimal modulation and coding scheme(MCS)which is suitable for the current channel conditions to transmit data,effectively adapting to the dynamic changes of channel and enhancing the system data transmission rate.In LTE uplink,AMC generally includes three steps,firstly,channel quality measurement is needed to obtain the CSI,and then MCS is selected according to CSI and fedback to the transmitter for the next transmission.In AMC,apparently channel quality measurement is the basis of the selection of MCS,and the performance of channel quality measurement directly affects that performance of AMC.SINR(Single to Interference plus Noise Ratio,SINR)is a common channel quality measurement metric,but the error of SINR estimation will lead to an unideal selection of MCS,which will cause the wrong configuration of transmission parameters and affect the system performance,therefore,the accuracy of channel quality measurement and the correctness of the selected MCS is important for system performance improvement.To enhance the accuracy of channel quality measurement in AMC,this paper studies the channel quality measurement algorithms in LTE system;and according to the development trend of AMC technology,the concept of machine learning is introduced into the LTE uplink AMC system and research on the concept of using machine learning algorithms to improve system performance is also studied.In the aspect of channel quality measurement,this paper focuses on two typical LTE scenarios,namely,the edge scene and the burst scene,in which the characteristics of channel quality are analyzed deeply.The common channel quality measurement algorithms based on multi-carrier are introduced,such as Exponential Effective SINR mapping(EESM)algorithm,Logarithmic Effective SINR mapping(LESM)algorithm and Harmonic Mean(HARM-MEAN)algorithm.The problems of existing channel quality measurement algorithms are studied,then from the aspect of single carrier,the channel quality measurement algorithms based on Turbo receiver and EVM are presented.On the basis of these algorithms,two improved algorithms are proposed to reduce the impact of too few samples of data on the accuracy of channel quality measurement.The Matlab simulation platform of LTE uplink adaptive system is built.Relying on this simulation platform,extensive simulations involved with the two proposals are conducted and their performance advantages are verified consequently.According to the development trend of AMC technology,the one-dimensional channel quality measurement metric cannot reflect the channel state deeply and the performance of this scheme is largely dependent on the specific model of the channel,this paper introduces the concept of machine learning to LTE uplink in AMC.This paper studies the problem of effective SINR mapping in channel quality measurement by using the idea of machine learning.Two typical machine learning algorithms are analyzed and introduced into LTE uplink AMC system,the AMC algorithm based on k-Nearest Neighbor(k-NN)supervised learning and Support Vector Machine(SVM).Given the characteristic that machine learning can adapt to the changing of environment,the ideal of using depth learning(Q learning)to control the MCS selection strategy is studied and the AMC algorithm based on Q learning is presented,further enhancing the performance of LTE uplink AMC system.Finally,the performance of the above machine learning algorithms have been verified by simulation in LTE edge scene.