Research On Pilot Design And Channel Estimation Based On Massive MIMO

Massive MIMO is one of the key technologies of 5G and it has the advantages of strong resolution,low transmission power and strong anti-interference ability.However,the massive MIMO system also has serious pilot pollution problems.When the massive MIMO system is adopted,the cell station can receive pilot signals from other cells.Because the pilot signal is responsible for signal channel estimation in the uplink,the pilot pollution problem can also affect the accuracy of the channel estimation.Pilot design and channel estimation in the case of pilot pollution in massive MIMO are researched and modified in the thesis.(1)It studies and proves that when the even-numbered ZC(Zadoff-Chu)sequence satisfies the condition that the sum of the cyclic shifts is odd,the cross-correlation modulus is close to zero.By using this conclusion,a pilot scheme different from other pilot sequence with odd-numbered ZC is designed.In order to satisfy the condition that the sum of the cyclic shifts is odd,four-type cell classification method is proposed,which can not only distinguish effectively the signals of this cell and others,but also increase the multiplexing distance of same pilot frequency among cells that can decrease the pilot pollution.(2)There is a big deviation in the case of pilot pollution of the traditional LS channel estimation and MMSE channel estimation,utilizing the designed pilot to have channel estimation proposed by the thesis that can decrease the estimation mean square error.At the same time according to the cross-correlation modulus is near to zero of the designed even-numbered pilot,the pilot channel estimation result can be obtained only by correlating the received signal with the known pilot when the pilot channel is estimated at the receiving end,which decrease the complicity of channel estimation.(3)To the problem of semi-blind channel estimation based on EVD method requires extra pilot overhead,an improved SVD semi-blind channel estimation scheme is proposed in the thesis.Combining the results of pilot channel estimation,the SVD decompositions are performed on the autocorrelation matrixes of the received signal and the transmitted signal,and then to have semi-blind channel estimation,which can improve the estimation accuracy.Even-numbered ZC code pilot sequences are used in conjunction with the proposed four-type cell classification method that can reduce the impact of pilot pollution while screening cell signals.Based on the proposed pilot sequence,ZC pilot channel estimation method proposed in the thesis is not only having high accuracy but also convenient and fast.To solve the fuzzy matrix problem in semi-blind channel estimation with traditional EVD method,the modified SVD semi-blind channel estimation method proposed in the thesis improves the estimation accuracy and overall system efficiency.