Low Complexity Channel Estimation Method for OFDM System in Doubly Selective Channel

牛　少博 (1251132)

Orthogonal Frequency Division Multiplexing (OFDM) is an efficient technique capable of establishing wideband digital communications by making use of multi- carrier transmission technology. Because of its high rate transmission capability and great immunity to multipath fading, OFDM has been widely adopted in vari- ous communication systems, such as Wireless Local Area Network (WLAN) stan- dard 802.11n, Integrated Services Digital Broadcasting for Terrestrial (ISDB-T) standard and also applied in the next generation mobile communication standard Long Term Evolution-Advanced (LTE-A). However, the mobile radio channel is always characterized as frequency-selective and time-varying channel. In such this channel the Channel State Information (CSI) changes rapidly. And the fast fading causes a serious performance degradation. Thus, a dynamic channel esti- mation is necessary for OFDM system.

The conventional 2 Dimensional Compressive Sensing (2DCS) based channel estimation method proposed gives a significant improvement of Bit Error Rate (BER) perfor- mance against the traditional methods, such as Minimum Mean Square Error (MMSE) and Least Square (LS). However, due to the large size of the measure- ment matrix, the high computational complexity is involved and degrades the system efficiency.

In this thesis we propose the 2-Step Compressive Sensing (2SCS) based chan- nel estimation method to reduce the computational complexity for OFDM via frequency-time doubly-selective fading channel. This method uses 2 types of small size of measurement matrices to determine the CSI in time and frequency domain sequentially. The proposed method achieves an considerable reduction of computational complexity while provides a reliable channel estimation performance. Using this proposed method the computational cost required for channel estimation process of OFDM via time-frequency doubly selective fading channel is reduced to 17% compared with conventional 2DCS method while keeping the BER performance at same level.