在无线通信的诸多场景，如深空通信和隐蔽通信中，都存在着链路余量不足、发射功率受限的问题，需要降低系统的接收门限，来提升链路的传输可靠性。降低接收门限的关键点有两个：一是采用高性能信道编码，降低译码门限；二是采用低信噪比同步解调技术，确保解调门限不高于译码门限。其中，前者通过合适的码字设计，已可充分逼近香农极限。而后者则存在技术瓶颈，需要不断增加导频来适配编码性能，而这又降低了系统的频谱效率。对此，国际上先进的做法是将译码与解调同步联合执行，利用译码反馈改善同步解调，修正后的解调结果再进一步改善译码，这种交替执行同步解调和译码的方法被称作迭代接收。迭代接收通过反馈机制，使信道编码增益作用于信号接收全过程，系统的接收门限仅取决于译码门限。然而，在现有的研究实践中，这一技术还存在着分析框架不完整、应用范围受限、工程落地难等问题。 本文在典型算法基础上，首先对迭代接收的收敛性进行分析。迭代接收的基本原理是基于译码信息来辅助信号的同步解调，本文基于估计理论中的变分推断框架，对典型算法实现方案予以统一分析。在此基础上，给出了算法的收敛精度和收敛条件，完善了迭代接收的理论依据。 其次，根据迭代接收需以帧同步为前提的特点，设计低信噪比信号并行帧同步方案。帧同步器通过并行方式克服载波和定时误差，并基于理论性能指标选取最佳的检测门限和状态机参数，可保障迭代接收前级模块的稳定运行。 接着，为克服迭代接收收敛范围窄、动态信号跟踪能力差的问题，提出收敛范围拓展算法。基于本文分析得到的目标函数计算公式，通过自适应网格搜索找到迭代接收算法合适的初始化位置，可有效避免局部收敛。此外，还考虑了一种候选列表机制，以减少收敛范围拓展所引入的计算开销。 最后，针对迭代接收算法需多次执行译码，计算复杂度高、数据吞吐率低的问题，研究迭代接收复杂度优化方案。基于迭代接收目标函数的形态重复性提出一种查表优化策略，可在计算复杂度和性能上取得合适折中。与理想同步条件下的译码过程相比，优化后迭代接收产生的额外计算开销约占比7%,使得同步不再成为接收过程中计算开销的最大来源。
In many wireless communication scenarios, such as deep space communication and covert communication, there are problems of insufficient link margin and limited signal power, which require lowering the reception threshold of the system and improving the data transmission rate. There are two key points to reduce the reception threshold: one is to use high-performance channel coding to minimize the decoding threshold; the other is to use low signal-to-noise ratio demodulation and synchronization technology to ensure that the demodulation threshold is not higher than the decoding threshold. Among them, the former can be sufficiently approached to the Shannon limit by suitable code word design. Conversely, the latter has a technical bottleneck and needs to continuously increase the pilot sequences to adapt to the coding performance. In this regard, the advanced international practice is to jointly execute decoding and demodulation synchronization, using decoding feedback to improve synchronization and the corrected synchronization results to further improve decoding. This alternate execution of demodulation synchronization and decoding method is called iterative reception. Through the feedback mechanism, the coding gain is applied to the whole signal reception process, and the system‘s reception threshold only depends on the decoding threshold. However, this technique still needs to be mature in existing research practice, and there are problems with an incomplete analytical framework, restricted application scope, and difficulty in engineering implementation. Based on the existing algorithms, the convergence of iterative reception is analyzed first. The basic principle of Iterative Receiving is to optimize the synchronization parameters of signals based on decoded information. Based on the variational inference framework in estimation theory, this paper makes a unified analysis of typical algorithm implementation schemes. On this basis, the convergence accuracy and conditions of the algorithm are evaluated to improve the theoretical basis for iterative reception. Secondly, according to the characteristics of iterative reception which requires frame synchronization, a parallel frame synchronization scheme for low SNR signals is designed. The frame synchronizer overcomes the carrier and timing errors in a parallel way, and selects the optimal threshold and state machine parameters based on four performance indicators to guarantee the stable pre-stage module of the iterative reception. Then, to overcome the problems of narrow convergence range of iterative reception and poor dynamic signal tracking capability, the iterative reception convergence range expansion algorithm is proposed. Based on the objective function formula obtained from the analysis of this paper, the iterative reception algorithm is initialized by adaptive grid search and interpolation to avoid local convergence. Finally, the iterative reception algorithm is studied for the problems of high computational complexity and low data throughput rate, which require multiple executions of decoding, and the iterative reception complexity optimization scheme is investigated. A map method optimization strategy is proposed based on the morphological repetition of the iterative reception objective function, which can achieve a suitable compromise in computational complexity and performance. Compared with the decoding process under ideal synchronization conditions, the additional computational overhead generated by the optimized iterative reception accounts for about 7%, so that synchronization is no longer the largest source of computational overhead in the reception process.
