Research On Channel Estimation In Wireless OFDM System

With the rapid growth of Digital communication in recent years, orthogonal frequency division multiplexing (OFDM) has received much attention as a promising scheme for high data transmission in wireless communication. In coherent detection OFDM systems, accurate estimation on channel information plays an important role. Basing on the analysis both of pilot assisted and blind channel estimation, this paper propose a new channel estimation algorithm which combines the wavelet and Singular Decomposition Value (SVD) low rank approximation to achieve a superior performance. At the same time, a frequency domain subspace algorithm is presented to overcome the huge computing cost in time domain algorithm, additionally; the method is insensitive to SNR changing.The dissertation is organized as follows:1. Key techniques and main advantageous and disadvantageous of OFDM are studied at the first, and then numbers of fading are analyzed to address the influence on channel estimation, Rayleigh fading simulation is presented as well.2. LS as well as MMSE algorithm and its application in OFDM channel estimation are studied. For block pattern, LMMSE algorithm is proposed for simpleness comparing with MMSE; in comb pattern channel estimation, some common interpolation methods are studied; in diamond pattern estimation, pilot spacing selection and Weiner filtering as well as its simplified style are analyzed. For a given pilot density, the author optimally analyses the pilot pattern so as to minimize the mean squared error (MSE) of the channel estimate with the use of a general interpolator.3. Following above, a new method based on wavelet denoising and low rank approximation MMSE channel estimation algorithm is presented. Using the wavelet denoising property, the original outcome after LS estimation is processed by wavelet to depress noise, and then the processed outcome is approximated by SVD. This new method can not only reduces the floor error but also can apparently reduce computing cost. In the end, simulation results are addressed.4. Aiming at the researching advances, a frequency domain subspace method is adopted to improve the performance of the time domain one. Comparing the exist ones, the estimating matrix in the method is not only simpler but also insensitive to SNR.