面向深空探测图像传输的全息压缩及无速率编码

In deep space exploring, the high-quality, high-efficiency, and high reliability transmission of the valuable images is in face of many restrictions, including the asymmetry of node ability/resource, the complexity and time-varying property of deep space channel, the large attenuation, the long propagation time, the asymmetry of the uplink and downlink. To break through these restrictions, the compressed sensing (CS) theory and Spinal code are brought into the image transmission system design, on considering that the traditional image compression and channel coding technologies have many limitations. Based on CS, full-information image compression technology is researched, which can achieve high-efficiency and high-quality compression with extremely-low complexity. And meanwhile, the compression technology has the characteristics of progressive and scalable decoding, robust to erasure and flexible coupling of decoding-coding. Based on the capacity achieving Spinal code, research is conducted on rateless code design for deep space channel. The designed rateless code can achieve high gain with low implementation complexity, so that the transmission ability of deep space channel is fully explored. Lastly, the CS compression, Spinal code, and the DTN-LTP protocol are combined to form a cross-layer transmission mechanism, which serves as an integrated multi-hop transmission scheme for high-efficiency and high-reliability transmission. To sum up, the difficulties of deep space communication are fully considered in this project, and the research is expected to provide a novel solution for future deep space image transmission, where much higher transmission quality and longer transmission distance will surely be required.

深空探测中，珍贵探测图像数据的高质、高效、高可靠传输面临节点能力/资源非对称、信道复杂时变、损耗大、时延长、上下行链路非对称等多重制约。为突破这些制约，并考虑到传统图像压缩及信道编译码技术的局限性，本项目引入压缩感知理论与Spinal无速率码进行探测图像传输系统的设计。基于压缩感知理论，研究具有极低编码复杂度的探测图像全息压缩技术，保证高效高质量压缩的同时，具有渐进可伸缩、可纠删、解码-编码灵活耦合特点；以近容量限传输的Spinal码为核心，研究复杂度低、增益高的深空信道无速率编译码技术，充分挖掘深空动态信道的底层传输能力；最后，结合压缩感知纠删、Spinal无速率码与DTN-LTP协议，设计跨层联合优化传输机制，为深空中继多跳通信模式下的高效高可靠图像传输提出综合解决方案。本项目充分考虑深空通信场景的技术难点，可为未来更高需求、更远距离的深空探测图像的压缩与传输提供可行的新技术途径。

为突破深空探测图像传输所面临的信道复杂时变、损耗大、时延长、上下行链路非对称、发收节点资源非对称等多重制约，本项目基于压缩感知理论与Spinal无速率码研究探测图像的压缩与信道编码技术，并通过与DTN协议栈下LTP协议的跨层结合，设计出联合优化传输机制，为深空通信中高效高可靠图像传输提出综合解决方案。截止2017年底，各研究计划要点全部执行，预期目标基本达到。..项目所取得的主要成果包括：(1) 以低复杂度探测图像压缩为目标，将分离压缩、分块压缩应用于CS图像压缩的编码端，并提出了相应的“分离-合并”图像解压法和基于重叠法的分块效应缓解方法。(2) 以高效高性能图像解压缩为目标，综合考虑探测图像在空间域、变换域所呈现的结构化先验、统计先验特征，提出了多域多先验联合利用的图像CS重构算法（JLMMP-CS）。(3) 以高性能连续探测图像传输为目标，基于CS编码的压缩与纠删一体特性，研究了对关键帧和非关键帧图像不等保护的分布式视频压缩感知（DCVS）系统，同时，为关键帧图像的CS重构提出了一种基于双密度双树复小波（DDDT-CWT）基的结构化CS图像重构算法，为非关键帧重构设计了一种采用双密度双树复小波变换作为双重稀疏模型的核心字典。(4) 深入研究了Spinal码在应用于深空通信场景时的编解码参数优化、工作模式选择等问题，并建立了Spinal码译码成功CDF模型。(5) 以降低Spinal码的译码复杂度为目标，提出了滑块反馈译码，通过在译码流程中结合使用滑动窗口和反馈确认，在不损失码率性能的前提下大幅缩短了Spinal码的译码时间。(6) 以综合实现协议栈框架下深空探测图像数据的高效可靠回传为目标，对CS、Spinal、LTP进行跨层结合，提出了基于上层纠删、基于上层容错两种跨层联合传输方案；以吞吐量最大化为目标、以满足最低重构质量为约束，为两种方案分别建立了优化模型并求解出了最优传输参数配置策略。(7)搭建了一套地火通信半实物仿真平台，完成了对所提深空探测图像跨层联合传输方案的综合评估验证。.