块衰落信道下无CSIT干扰对齐的理论与算法研究

Interference alignment (IA) aims to enhance the degrees of freedom (DoF) of interference channels by controlling the impairment effect of interference, known as a new possible solution for interference management. Primitively, the implementation of IA needs time/symbol extension associated with the assumption of ideal time-variant channel model, as well as the complete channel state information at transmitters (CSIT), therefore it is extremely hard to be fulfilled in practical systems. Although IA with limited CSIT and IA with delayed CSIT can clear a few implementation barriers, IA without CSIT is more promising due to its easy but efficient precoding design. However, the current research only verifies the feasibility of IA without CSIT for some ideal time-variant channels with unusual features, which limits its practical value. This project puts the feasibility problem of IA without CSIT into the scope of general block fading channel model with less ideal features, and intends to systematically solve several key implementation problems of IA without CSIT. Particularly, for multi-user MISO broadcast channel (BC), multi-user MxK X channel and multi-user SISO interference channel (IC), provided that the transmitters only have the knowledge of the coherence times involved, this project will (1) identify the sufficient condition or the necessary and sufficient condition for the feasibility of IA without CSIT in the aforementioned block fading channel model and give explicit implementation algorithm; (2) derive the transmission performance of IA without CSIT given the block fading model with or without memory; (3) solve the user selection problem and the scheduling problem associated with IA without CSIT when a great amount of users are available. This project will facilitate the development of the theory of IA and accelerate its application.

干扰对齐技术以控制链路间的干扰提高系统的自由度为目标，尝试解决同频干扰问题，但是干扰对齐通常需要采用理想时变信道假设下的时域扩展、且要求发送端已知完全CSIT，在实际系统中难以实现。部分CSIT和滞后CSIT干扰对齐可以一定程度解决上述问题，但相比之下无CSIT干扰对齐实用性更强。然而目前无CSIT干扰对齐的可行性仅在少量有特殊结构的理想时变信道得到证实，对应用的指导有限。本项目拟全面考虑非理想的块衰落信道模型，系统地解决无CSIT干扰对齐的可行性等若干实际问题。具体针对三类多用户信道拓扑：MISO广播信道、MxK X信道和SISO 干扰信道，在仅假设发送端已知信道相干时间的条件下，解决块衰落模型下无CSIT干扰对齐的可行性问题及实现算法；推导无记忆和有记忆块衰落模型下无CSIT干扰对齐的传输性能；并分析多用户网络下采用无CSIT干扰对齐的用户调度。本项目将完善干扰对齐的理论并推动其应用。

本项目主要研究盲干扰对齐在应用中的关键技术。为获得盲干扰对齐，目前常用的方法包括1）根据块衰落的相干特性来设计盲干扰对齐，2）采用接收端配置多模天线通过调节天线转态来调整信道结构使得盲干扰对齐可以实现。本项目研究发现上述两种方法在现实场景中需要配合使用。最主要的研究成果是，多模天线的切换方式会影响系统总的切换频次，而通过有效设计可以获得相同盲干扰对齐自由度的性能下降低总切换频次，进而获得更好的接收端信噪比。针对多用户MISO干扰信道，我们提出的S-BIA算法能实现最少的天线切换频次。本项目同时也研究WSN网络中的协作MAC机制，并对多种协作机制进行性能分析；也研究WSN网络中的数据收集策略，提出了能量有效的基于压缩感知方法的数据收集策略。