Research On Advanced Receiver Of Non-orthogonal Multi-carrier Modulation

To meet the challenge of significant and continuing growth of mobile traffic and wireless-connected devices,the fifth-generation(5G)mobile communications system has got extensive attention and research by enterprises,research institutes and universities of the world at present.As one of the key techniques of 5G,new waveform is an essential prerequisite to network planning and equipment development.One of the main drivers for new waveforms is to enable efficient multiple access in the light of the requirements of future 5G wireless communication systems which will have to handle a very diverse variety of scenarios,including enhanced mobile broadband(eMBB),massive machine type of communication(mMTC)and ultra reliable and low latency communication(URLLC).Recently,the filtered waveform schemes,such as filterbank mul-ticarrier(FBMC),generalized frequency division multiplexing(GFDM),have received much attention as candidates for 5G waveforms.However,the non-orthogonality of the filtered waveform systems incurs se-vere loss in performance.In this thesis,the advanced receiver design of the non-orthogonal multi-carrier modulation techniques will be investigated.Firstly,on the basis of factor graph(FG),we describe the update rules of the sum-product algorithm in detail and study the approximate inference algorithms of bayesian network to establish the general concept of message passing.We then focus on the presentation of the approximate message passing(AMP)related information.Under some reasonable assumptions,a detailed derivation of the AMP algorithm is presented,which makes quadratic Gaussian approximation of the standard message passing algorithm in the large system limit.In order to resolve the problem of the limitation of AMP for the ill-conditioned matrix,orthogonal AMP(OAMP)is introduced and the iterative steps are presented,which provide a sound theoretical basis for the following research on the receiver design of non-orthogonal multicarrier system.Subsequently,we address the problem of the non-orthogonality in time and frequency of GFDM sys-tern especially with multiple-input multiple-output(MIMO)and propose a highly efficient iterative detection algorithm for MIMO-GFDM.We first review the principle of GFDM modulation both in time and frequen-cy domain briefly and reveal that the non-orthogonality of the adjacent subcarriers and time slots engenders performance degradation in bit error rate(BER)using traditional linear receivers.To eliminate this degrada-tion,we propose an optimal iterative algorithm for GFDM based on OAMP algorithm.We study the special structure of the equivalent channel matrix,investigate the block-diagonalization of that and derive the exact expression of the diagonalization of submatrices,which is utilized in MIMO-GFDM system to transform the equivalent channel matrix into block-tridiagonal structure.We further propose the use of the conjugate gra-dient(CG)method in matrix inversion instead of direct calculation,dinimishing the complexity incurred by matrix inversion per iteration of OAMP.Exploiting the CG method and the traits of block-tridiagonal matrix,we develop the CG-OAMP to estimate symbols in MIMO-GFDM systems.According to the thoroughly e-valuation of the computational complexity of the CG-OAMP,the introduction of CG method can significantly reduce the cost engendered by the matrix inverse operations in OAMP.Numerical simulations indicate that the proposed algorithm performs closely similar to the optimal performance at reduced computational costs in MIMO-GFDM signal detection.Finally,we focus on the circular-filterbank multicarrier(C-FBMC)and come up with an iterative re-ceiver algorithm on the basis on OAMP.We first study the C-FBMC modulation matrix,deriving the real-orthogonality of the columns and the orthogonality of the rows of the modulation matrix which is used for directly reducing the complexity costed by matrix inversion from being cubic to be linearithmic with the num-ber of subsymbols.Combining the OAMP algorithm with these interesting properties,the advanced iterative detector is proposed for single-input single-output(SISO)C-FBMC system.It is then extended to MIMO system.In order to reduce the complexity incurred by the matrix inversion,we propose to transform the e-quivalent channel matrix into block-diagonal matrix.Combining the characteristics of block-diagonal matrix and the traits of the C-FBMC system in frequency domain,we present a OAMP-based algorithm for MIMO-C-FBMC signal detection.Simulation results show that the proposed iterative receiver can achieve optimal performance at diminished computational cost,which demonstrate its high-efficiency and low-complexity.