Abstract

In this paper, we investigate blind channel estimation in cyclic prefixed orthogonal frequency division multiplexing (OFDM) systems using precoding matrices. Many studies show how the performance of bit error rate (BER) and mean square error of channel estimation are impacted by using a circulant matrix as a precoding matrix. We investigate a new structure for precoding matrix in which the first row is generated by an exponentially decreasing equation, and the Toeplitz final precoding matrix is generated via Singular Value decomposition (SVD). This type of SVD-based precoding matrices is been simulated and a modification is presented to improve the system performance. The resulting precoding matrix is employed in two different OFDM systems to estimate the channel frequency response vector up to scalar ambiguity. The proof of the full rank property is presented along with other criteria that help with the development of various precoding matrices