Research Of Pilot-Aided Sparse Channel Estimation In Ofdm System

Orthogonal frequency division multiplexing (OFDM) technique, as one of the most popular mobile communication technologies, has strong ability to resist frequency selective fading and narrowband interference, high spectrum efficiency. Also, it takes the advantage of digial implementation including fast Fourier transform (FFT) and inverse FFT (IFFT). As a result of these factors, OFDM receives extensive attention from researchers. Channel estimation, as the basis of related detection, demodulation and equalization, greatly affects the overall performance of OFDM system.In wireless communication systems, most multipaths are sparse, which means the channel is composed of several important paths. This sparsity is more obvious in high-speed communications and broadband wireless communication systems. The traditional channel estimation methods ignore the sparsity which leads to unsatisfactory estimation accuracy and efficiency. Based on this situation, this paper focuses on the estimation of sparse channels:how to use fewer pilot symbols implement channel estimation and improve estimation accuracy.Cyclic delay diversity (CDD) channel is a special form of sparse channel. The traditional uniform pilot may all cyclically fall on the points where the values of the channel frequency response are same, leading to severe channel estimation error. To avoid this problem, this paper proposes a non-uniform pilot based channel estimation for CDD-OFDM systems and improves the accuracy of channel estimation. This paper also compares the performances of different scattered data interpolation algorithms in the channel estimatioiIn addition, this paper proposes to apply compressive sensing based channel estimation toCDD-OFDM systems with uneven pilots. By using the uneven pilots and COSaMp estiation method, we can estimate the position and values of significant taps. This method can improve channel estimation accuracy and also reduce the number of necessary pilots and thus improve the utilization rate of the channel.For time-varying channels, basis expansion model (BEM) is usually used to model channel and implement channel estimation. By using inverse method, we can estimate the coefficients of basis function based on the estimated Fourier coefficients, which reduces the complexity of BEM coefficient estimation. However, the inverse method ignores the estimation error of Fourier coefficients which leads to the inaccuracy of channel estimation. To solve this problem, this paper proposes a LMMSE-based model to improve the inverse method for channel tap reconstruction. In addition, this paper applys this improved BEM estimation method to sparse channels, which reduces the complexity of computation and also improves the spectral efficiency.