A Study on Interference Suppression and Frequency Offset Estimation for OFDMA Uplink Systems

• Main Authors: Tzu-chien Chuang, 莊子建
• Other Authors: Ju-ya Chen
• Format: Others
• Language: zh-TW
• Published: 2009
• Online Access: http://ndltd.ncl.edu.tw/handle/ew46h6

碩士 === 國立中山大學 === 通訊工程研究所 === 97 === Orthogonal Frequency Division Multiple Access (OFDMA) uplink systems have been adopted generally in wireless communication in recent years. By combining Orthogonal Frequency Division Multiplexing (OFDM) with Frequency Division Multiple Access (FDMA), OFDMA systems divide all subcarriers in frequency domain into several mutually exclusive subbands, and assign the subbands to different user, transmitting different kinds of data by the orthogonality between subcarriers. The users’ Carrier Frequency Offsets (CFO) caused by Doppler effects and/or misadjusted local oscillators may destroy the orthogonality between subcarriers, then lead to Multiple Access Interference (MAI) for all users, the MAI may degrade the performance of bit error rate. Several MAI suppression or cancellation schemes have been investigated to mitigate the effect of MAI, however, these schemes work on the hypothesis of known CFOs. As the result, it is the crucial issue for the multiuser CFOs estimation in OFDMA uplink systems. In this paper, we propose the Minimum Mean Square Error Criterion (MMSEC) estimation scheme to achieve multiuser CFOs estimation in Additive White Gaussian Noise (AWGN) environment. By modifying the MAI suppression weighting in [9], the MMSEC scheme search desired user’s true CFO with CFO trial parameter, while the CFO trial parameter is equal to true CFO, the modified MAI suppression weighting will suppress the MAI ideally and then separate the desired user’s estimated signal from received signal. Thus the distribution of estimated signal is highly concentrated in each of quadrant in the signal constellation, and it will have the minimum mean square error in each of quadrant. Compare with the MUtiple SIgnal Classification (MUSIC) scheme [12] and the Estimation of Signal Parameter via Rotational Invariance Technique (ESPRIT) scheme [13], they deal with the fully loaded problem by extending cyclic prefix, that may cause the redundant power consumption or degrade data throughput, the advantage of the MMSEC scheme is that it can works on fully load without extending cyclic prefix. Because despite fully loaded, the CFO trial parameter search desired user’s true CFO in reasonable range until it causes minimum mean square error. The simulation result shows that the non-fully loaded system mean square error (MSE) performance is superior to the MUSIC and ESPRIT schees at high signal to noise ratio, while system is fully loaded, the system MSE performance is also acceptable.