Research On Faster-than-nyquist Signaling

To achieve the stringent spectral efficiency requirement of further communication networks,potential communication technologies that can enhance the spectral efficiency,such as faster-than-Nyquist(FTN)signaling,has attracted substantial attention.FTN signaling is a classic non-orthogonal signaling format that has been proved to have a greater Shannon capacity compared to conventional Nyquist signaling with practical shaping pulses.The capacity advantage of FTN signaling comes from the higher symbol rate,which in return inevitably causes severe intersymbol interference(ISI).Based on the characteristic of FTN signaling,we conducted the research on the reduced-complexity detection methods and code design for FTN systems.The main contributions of this dissertation are summarized as follows:We proposed two novel M-BCJR algorithms for FTN signaling based on the Ungerboeck channel observation model.By taking account of the influence from the “future” symbols,the proposed algorithms determine the most reliable trellis states according to a different metric.We proved that the correct trellis states can be remained in the reduced trellis for both proposed algorithms thereby providing a better error performance compared to the conventional M-BCJR algorithm based on the Ungerboeck channel observation model.Simulation results verifies our analysis and demonstrate a good trade-off between the error performance and detection complexity.We analyzed two commonly considered multi-layer FTN transmission structures that enable reduced-complexity detection.In specific,we discussed their transceiver structure designs and the application of successive interference cancellation(SIC).Furthermore,according to their transceiver structure,we proposed two power allocation strategies.We also derived the capacity of the two structures.Interestingly,we showed that one structure can obtain the degree-of-freedom(Do F)advantages of FTN signaling while the other one can only obtain the signal-to-noise ratio(SNR)advantages.Those results can serve as a guideline for practical FTN designs.Meanwhile,our analysis are explicitly verified by simulation results.We proposed a novel code based channel shortening(CCS)algorithm for detecting FTN signals.The proposed algorithm applies a special type of convolutional codes to absorb the channel memory induced by FTN signaling,where the idea of CCS algorithm can also be extended to the equalization for general ISI channels.We call this type of convolutional codes the output-retainable convolutional codes(ORCCs),whose previous code symbols can be retained by the current code trellis state.Based on the property of ORCCs,we introduce the CCS algorithm.We derived the error performance bounds of the CCS algorithm.According to the error performance bounds,we further proposed a novel code search algorithm to find good ORCCs.Simulation results demonstrate that the proposed CCS algorithm can have a better error performance compared to conventional channel shortening algorithms.To enhance the error performance of FTN systems,we proposed various channel coding schemes for FTN signaling,include Turbo codes,serial-concatenated convolutional codes,and self-concatenated convolutional codes.We use both the distance spectrum and EXIT chart as tools for our code design.For the Turbo code,we alter the receiver structure to improve the error performance of Turbo coded FTN signaling.Furthermore,for concatenated convolutional codes,we derived the corresponding EXIT characteristics and optimized the coding structure.Our simulation results show superior error performance of the proposed coding schemes compared to the Nyquist counterpart,which also agrees with our EXIT chart analysis.According to the main contribution of the study,we also summarized the future research directions for FTN signaling at the end of this dissertation.