Research On Single Channel Blind Separation Of Paired Carrier Multiple Access Signals

New frequency reuse technology of Paired Carrier Multiple Access(PCMA)can not only double the spectrum efficiency of the satellite channel,but also improve the anti-interception capability of the communication signals.Both communication terminals of PCMA use the same uplink and downlink resources,and each terminal receives the mixture of two signals overlapped completely in both time and frequency domains.Regarding uncooperative reception scenarios,the third part is concerned with information on both sides of PCMA,but without knowing any prior knowledge.Two signals need to be recovered from the co-frequency mixture received by single channel receivers,which belongs to the issue of single channel blind separation(SCBS).This paper mainly researches on SCBS of PCMA signals while laying great emphasis on the separability based on information theory,high precision parameter estimation,low complexity per-survivor processing(PSP)blind separation and multi-level joint iterative processing combined with coding.The main assignments and innovations are summarized as follows.1.The separability of PCMA signals is studied from the perspective of information theory.The mixture signal model is regarded as a multiple-access channel with memory.For different channel memory length,the forward and backward sum-product recursion of the BCJR algorithm and the Monte Carlo integration-based method are proposed to compute the information rate of the mixture signals.The separation performance lower bound is obtained via Fano inequality.The influence of the modulation parameters on the information rate bound and separation performance bound is also analyzed.The resulting information rate bound and separation performance bound provide the theoretical reference for the separability of the mixture signal.2.Several high precision parameter estimation algorithms are proposed,and the Cramer-Rao bounds are derived.Parameter estimation of the mixture signals is indispensable.In order to improve the separation performance,the high precision parameter estimation algorithms based on oversampling and data-aided(DA)are researched.The oversampling fourth method,the DA maximum likelihood(ML)amplitude estimation algorithm,the decisiondirected ML phase estimation algorithm,the maximum mean fourth method and the DA-ML timing estimation algorithm are proposed respectively.In order to evaluate the parameters estimation accuracy,the Cramer-Rao bounds(CRBs)for constant scalar and vector parameters,the modified Cramer-Rao bounds(MCRBs)for separate and joint estimation are derived.The high precision parameter estimation algorithms presented in this paper can attain or approach to the Cramer-Rao bounds.3.Two low complexity PSP separation algorithms are proposed from the channel and signal level.The complexity of the PSP SCBS algorithm increases exponentially with the channel memory length and the signal modulation order.How to effectively reduce the complexity is the bottleneck problem of that the PSP blind separation towards practical.From the perspective of the channel,the reduced-state decision feedback-feedforward PSP(RSDFF-PSP)SCBS algorithm is proposed for long channel memory(),in which the global decision feedback and delayed decision feedforward filters are designed to process the pre-and post-cursors.The proposed algorithm provides the good tradeoff between complexity and performance.From the perspective of the signal,the Ungerboeck set partitioning PSP(SP-PSP)SCBS algorithm is proposed for the higher-order mixture signals(8PSK/8QAM/16QAM),in which the constellations of the higher-order mixture signals are divided into several subsets combination.The proposed algorithm effectively reduces the alphabets size that further reduces the complexity,and achieves separation of the higher-order PCMA signals.4.Two low complexity joint iterative separation algorithms combined with different coding are proposed.The multilevel joint iterative processing model including parameter estimation,blind separation,demapping and decoding module is established through a combination of “BICM-ID scheme” and “Turbo principle”.The soft output PSP blind separation,soft in soft output(SISO)demapping and SISO decoding modules are researched.The decision feedback-feedforward PSP(DFF-PSP)iterative separation algorithm combined with convolution code is proposed.The reduced-state soft output PSP blind separation and Turbo Product Codes(TPC)iterative decoding structure are researched.The decision feedback and Ungerboeck partitioning PSP(DF-UP-PSP)iterative separation algorithm combined with TPC is proposed.The proposed algorithms can achieve the organic combination of low complexity PSP blind separation and error correction decoding at low signal-to-noise.