Research Of Channel Estimation In High Mobility OFDM Systems

OFDM is becoming widely used in wireless communications due to its high bandwidth efficiency and robustness with regard to multipath fading channel. It is considered to be one of the key technologies for next generation mobile communications, which can be applied in high mobility scenarios, such as high-speed highway communication. In high mobility OFDM systems, wireless channel becomes fast time-variant frequency selective channel. The Doppler spread caused by time variance will severely affect the performance of OFDM system, and at the same time destroy the orthogonality between OFDM subcarriers which will lead to a serious Inter-Carrier Interference. And ICI makes the channel estimation no long accurate. Therefore, channel estimation becomes a critical problem in high mobility OFDM systems.In this paper, we mainly study the channel estimation in high mobility OFDM systems. Firstly, we analyze the principals of OFDM technologies and propose the baseband system model. Then, we study the physical and statistical characteristics of the wireless channel, especially focus on analyzing large or small scale fading characteristics of the wireless channel and the ICI caused by Doppler spread,and establish a wireless channel model for high mobile communication scenarios. Secondly, we study the channel estimation methods in OFDM systems, especially research on pilot structures, channel estimation criteria. Finally we analyze the channel estimation performance by computer simulations. Regarding the fast time-variant frequency selective channel of high-speed OFDM system, we propose a low complexity iterative channel estimation method based on QR matrix decomposition algorithm).In this method, the detected symbols are feed back as additional pilot signal to refine channel estimation results, a QR matrix decomposition method is proposed to implement a low complexity channel estimation algorithm, and a weight coefficient is proposed to imitate the error propagation brought by the original pilot symbols. In the iterative processing, the pilot symbol matrix is decomposed by using QR matrix decomposition algorithm. Thus, the dimension of pilot symbol matrix is reduced, which makes the computation simple. As the simulation results show, the proposed channel estimation method converges faster than conventional channel estimation methods. When the Doppler spread comes to0.1, the BER performance of the proposed channel estimation algorithm has a gain of2-3dB SNR for the same BER, nearly achieve the perfect channel estimation.