Pilot Sequence Selection with Low Pilot Contamination in Downlink Massive MIMO-OFDM Systems

• Main Authors: Kang-Cheng Lin, 林剛正
• Other Authors: Chih-Peng Li
• Format: Others
• Language: zh-TW
• Published: 2015
• Online Access: http://ndltd.ncl.edu.tw/handle/b5c895

碩士 === 國立中山大學 === 通訊工程研究所 === 104 === In this paper, we consider a channel estimation problem which based on a multi- cell with downlink (DL) training framework for frequency duplex division (FDD) massive multi-input multi-output (MIMO) orthogonal frequency division multiplexing (OFDM) systems. In traditional MIMO-OFDM systems, transmitter transmits pilot sequences with different OFDM symbols which utilize time to satisfy orthogonality. In order to improve the performance of channel estimation, transmitter distributes larger power to OFDM symbol with pilot sequences in frequency domain. This procedure could cause the high peak-to-average power ratio (PAPR) effect in time domain. Using Zadoff and Chu (ZC) sequences as our pilot sequences which arrange them to both coherence bandwidth and coherence time interval to prevent high PAPR effect. According to the deviation of estimation performance, the MSE of LS estimator is dominated by pilot contamination which is proportional to the sum of cross-correlation of pilot sequences from inter-cell. We appropriately select pilot sequences with Cross Entropy (CE) algorithm could mitigate pilot contamination. According to the simulations, our scheme can mitigate the pilot contamination that further improve the estimation performance.