Research On OFDM System Channel Estimation By MB-LSF Algorithm

Orthogonal Frequency Division Multiplexing (OFDM) is viewed as a kind of multi-carrier digital modulation or multi-carrier multiplexing technology. This technology uses huge quantity of mutual orthogonal sub-carriers to transmit information, thus it can greatly improve the spectrum efficiency of communication system, and also reduce the interference brought by multi-path effect. So, OFDM became the core of the forth generation of wireless communication system.Channel estimation is a very important researching area in OFDM system, and it could be classified for two types, pilot aided channel estimations and blind channel estimation. The most classic and basic channel estimation method usually use Least Square (LS) algorithm, MMSE, or some modified MMSE algorithms to estimate the channel respond of pilot positions, and then get the whole channel respond via interpolating the CR on pilot positions.This thesis focused on Model-Based Least Square Fitting (MB-LSF) channel estimation algorithm. This method averagely divides channel into many blocks, and then use a suitable polynomial to approximate and fit a block of channel. This algorithm could be developed from LMMSE, and it doesn’t need any statistic information about the channel, but its performance is as good as LMMSE. After that, the concept of "pseudo-pilot" is introduced to assist MB-LSF algorithm. That is choosing data symbols of some certain positions to estimate its real symbols, and then use them as true pilots in MB-LSF algorithm. A new approach is introduced to decrease the complexity when researching the "pseudo-pilot", without decreasing the channel estimation performance.