Research On Detection And Channel Estimation Techniques For Non-orthogonal Modulations

Improving the spectrum efficiency has always been an important task for wireless communi-cations.In conventional orthogonal communication systems,the only way to improve spectrum efficiency is to adopt higher order modulation.However,the high-order modulation makes the system more sensitive to noise and non-linearity of the devices,which leads to significant perfor-mance degradation.Under such circumstances,we have to resort to non-orthogonal techniques,seeking to enhance spectrum efficiency fundamentally.FTN and SEFDM,as two non-orthogonal transmission techniques,have attracted much attention and are worth studying.Based on the analysis of basic principles of FTN and SEFDM,the dissertation focuses on the research of FTN and SEFDM receiver techniques with low complexity and good performance.The low-complexity factor graphed-based detection algorithms for FTN and SEFDM,and the channel estimation algorithms based on block-type pilots and comb-type pilots for SEFDM,have been researched.Firstly,the fundamentals of FTN and SEFDM are introduced.For FTN,the basic model is introduced and the high spectrum efficiency is analyzed through time-domain waveform as well as power spectral density.Then,the shaping filter and Mazo limit of FTN are studied.The ad-vantages of FTN,compared with orthogonal systems,are explained in aspect of system capacity.For SEFDM,the system model is introduced.The time-frequency view shows the advantages of SEFDM compared with OFDM.In addition,IFRFT-based and IDFT-based signal generation schemes are introduced,and the latter is the basis of the subsequent chapters.Then,low-complexity detection algorithms for FTN are studied.The dissertation proposes the FG-BP detection algorithm for FTN.According to the relationship among FTN symbols,the factor graph is established and the message passing in the BP algorithm is derived.The serial-schedule updating mechanism is also defined.In order to further reduce the complexity of the detection algorithm,GA is employed to model the interference and the noise together as a Gaussian noise.Then,the FG-BP algorithm is extended to FTN systems under multi-path fading channels and Turbo structure is applied to improve the BER performance.Through comparisons and analysis,it is demonstrated that the FG-BP detection algorithm for FTN can obtain remarkable performance with low complexity.Next,low-complexity detection algorithms for SEFDM are studied.Firstly,aiming at the problem of serious out-band spectrum leakage caused by the conventional CP,a novel continuous phase CP insertion scheme is proposed.With the aid of the continuous phase CP,both the phase continuity and the property of circular convolution can be guaranteed.Then,the factor graph of SEFDM is established and the FG-BP detection algorithm is derived both under the AWGN and multi-path fading channels.Since the factor graph of SEFDM is almost full-connected,an adaptive edge selection scheme is proposed,which can significantly reduce the detection complexity.Finally,the channel estimation for SEFDM is also investigated.The dissertation proposes the TCGTR channel estimation algorithm based on block-type pilots.The TCGTR algorithm can alleviate the ill-conditioning of the pilot matrix and improve the accuracy of channel estimation.However,channel estimation algorithms based on block-type pilots can be only employed in slow fading channels and static multipath fading channels due to the too much overhead.Consequently,a three-stage joint iterative channel estimation,detection and decoding algorithm based on comb-type pilots is proposed to better accommodate the rapid channel variation.Firstly,in the proposed algorithm,a partially orthogonal comb-type pilot is designed.Then,a novel channel estimation algorithm,using the feedback of the decoder for interference cancellation,is proposed.Joint it-eration is executed through exchanging information among the channel estimator,detector and decoder.With the aid of the three-stage Turbo structure,the accuracy of channel estimation and detection can be improved simultaneously.And the rapid variation of fast fading channels can also be better accommodated because of the well-designed comb-type pilot.