Signal Detection For Faster-than-nyquist(FTN) System With Large Compression Ratio

Spectral efficiency is a key objective in the design of a modern digital communication system.Faster-than-Nyquist(FTN)signaling can improve spectral efficiency by increasing the data transmission rate while preserving the bandwidth.However,FTN transmits symbols over Nyquist rate resulting in ISI inevitably.Moreover,with the increase of the compression ratio,the ISI introduced is more serious,limiting the use the optimal detection algorithms.Therefore,it is important to design sub-optimal algorithms,which achieve a proper tradeoff between the complexity and performance.The main research contents include:Detection for FTN system with small compression ratios is analyzed and studied.This thesis introduces the successive interference cancellation(SIC)algorithm and BCJR algorithm based on three observation models,realizing the iterative equalization for FTN with small compression ratios.In the BPSK modulated FTN system with a compression ratio of 50%,two detection algorithms achieve ideal performance at a SNR of 4 dB.For the problem of high complexity of BCJR algorithm in FTN system with large compression ratios,several sub-optimal detection algorithms are studied,including offset-BCJR,M-BCJR,M*-BCJR,and U-M-BCJR.In addition,the convergence processes are monitored by EXIT chart when above algorithms are used as a detector in an iterative loop.In the BPSK modulated FTN system with a compression ratio of 65%,several reduced-complexity BCJR algorithms studied in this thesis can detect FTN signals effectively.To this end,the backtracking M-BCJR algorithm,a new algorithm based on symbol tree,is designed to further reduce the complexity of BCJR algorithm.In this thesis,the symbol tree theory is applied to deriving the recursion for BCJR algorithm and in turn extending to M-BCJR algorithm.Then,considering the unstable LLR values generated by M-algorithm based detector,two backtracking-assistant algorithms based on symbol tree are introduced,including extended symbol tree and fitted symbol tree.Compared with other M-BCJR algorithms,the analysis shows that the proposed M-BCJR can reduce the overhead of recursive calculation and symbol tree expansion by half,thus reducing the complexity of the BCJR algorithm effectively.Finally,the new algorithm is assessed in the BPSK modulated FTN system with a compression ratio of 65%,of which the complexity is reduced by 50%~97% compared with other reduced-complexity BCJR algorithms,while keeping the same performance.