Ofdm System Channel Estimation Algorithm

Orthogonal Frequency Division Multiplexing (OFDM) is a parallel transmission scheme of Multi-Carrier's Modulation. It has a good capability of countering multi-path fading and high data transmission rate. So OFDM has been regarded as one of the dominating technologies of 4G wireless communications system.Channel estimation is one of the research hotspot. At present, based on the research productions all over the world, channel estimation can be generally separated into three kinds: pilot-aided channel estimation, blind channel estimation and semi-blind channel estimation. Pilot signals should be inserted into data signals in Pilot-aided channel estimation, pilots take up some spectrum, but this method is easy to realize. Blind channel estimation doesn't need pilot signals, has high spectrum utilization, but has complicated realization.This paper researched on channel estimation based on the comb pilot, and analyzed Least Square (LS) and Linear Minimum Mean Square Error (LMMSE) and Singular Value Decomposition - Linear Minimum Mean Square Error (SVD-LMMSE). And it adopted three interpolation methods, those were linear interpolation, second order interpolation and DFT based time-domain zero-inserting interpolation. When the channel estimation uses the Least Square rule, the method has a problem with the noise, so this paper used a wavelet transformation modulus maximum method to reduce the noise's influence. The MATLAB simulation results indicated that the wavelet transformation modulus maximum method could reduce the noise's influence effectively,and improve channel estimation's performance of the OFDM system.This paper studied the blind channel estimation based on the subspace method as well. The paper recommended the principle and conditions of the subspace method; the paper also introduced a normal application in which the length of Cyclic Prefix is longer than the order of the channel, in the method, noise subspace and signal subspace were both considered. Besides, the paper introduced a method in which the subspace method can also be applied, the length of the Cyclic Prefix could be shorter than the order of channel or even be zero, and the method narrated above is the Virtual Carriers technology. The MATLAB simulation results proved that when the length of Cyclic Prefix is shorter than the order of channel or even the length of Cyclic Prefix is zero, the Virtual Carriers technology still has favorable performance.