Jacket Transform And Its Application In Interference Cancellation For MIMO Systems

Abstract:Amidst numerous matrices that are being utilized in engineering applications, Hadamard matrices are in the forefront, which are used in signal sequence transforms, image analysis, signal processing and so on. The concept of Jacket matrix was conceived by Lee Moon Ho in1989, and in2000it was officially named as Jacket matrix which derived from the center weighted Hadamard matrix. It is the generalized extension of the well known Hadamard matrices, and the matrix addresses many problems in information and communication theories. Since the inverse matrix of Jacket matrix can be determined easily, Jacket matrix and its transforms have been extensively studied and widely applied in digital signal processing, wireless communications, cryptography, and so on.Currently, based upon the thought of the discrete fractional fourier transform (DFRFT), several fractional signal transforms have been proposed. However, the idea is rarely used in Jacket transform. The purpose of this thesis is to introduce the discrete fractional Jacket transform (DFRJT) with a fast algorithm based on the theory of DFRFT. According to the characteristics of Hadamard matrix, this thesis firstly puts forward a fast algorithm for DFRJT based on Hadamard matrix, and then proposes the generalized fast DFRJT with sparse matrix. The factorable FRJM of large size can be decomposed using the proposed fast algorithm into two ways. One is the Kronecker product of Hadamard matrix and FRJM of small size, the other is the Kronecker product of several FRJMs of small size. And then the DFRJT with eigen-decomposition is implemented. Compared with the direct computation approach, the proposed algorithm obviously decreases the computational complexity and increases the computational speed.Since the channel is broadcast, the transmitted information interfere each other in MIMO systems. In the thesis, a simple Interference Cancellation scheme based on eigenvectors of Jacket matrix is proposed for MIMO systems. In this scheme, eigenvectors of Jacket matrix are implemented at transmitter for pre-coding, so the signal has the same sign-change frequency as the eigenvector of Jacket matrix multiplied with it. Receivers find the desire signal out of interfering signal with the known sign-change frequency.Then minimum mean-square error (MMSE) decoding technique is implemented for removing the interferences. The proposed scheme has obvious low complexity, good performance and simple implementation.7figures,5tables,75references.