基于信息交互理论的无线信号的迭代估计与检测

【作者】 郭心悦；

【导师】 胡波；

【作者基本信息】 复旦大学 ， 电路与系统， 2008， 博士

【摘要】 无线信号检测作为无线通信系统中的关键问题,在下一代移动通信中面临很多新的挑战。迄今为止,不管是针对传统SISO系统或新兴MIMO系统,都已提出许多信号检测的算法,但是如何能在尽可能低的计算复杂度下获得最佳的检测性能,依然是研究的热点问题。本文在前人工作的基础上,以信息交互为理论框架,借助因素图模型为辅助工具,对无线信号的迭代检测技术进行了算法研究。首先,在理想信道估计的假设下,论文研究了SISO系统的迭代信号检测技术。通过对似然函数的分解,得到信号检测问题的因素图描述,在此基础上应用和积算法推导出信号检测的迭代算法。进一步,为降低和积算法的计算量,提出一种基于SP-PDA的快速迭代信号检测算法。仿真表明,相比最优检测算法,基于标准和积算法的迭代信号检测算法可以有效降低算法的复杂度,同时又可以获得逼近最优算法的性能;而快速迭代信号检测算法则在性能略有下降的情况下,进一步将计算复杂度从指数关系降低为线性关系,获得了性能与复杂度的折中。其次,信道均衡与信号检测最佳性能的获得依赖于准确的信道知识。针对未知且时变的信道参数的情况,本文研究了基于因素图的联合信道估计与信号检测的迭代算法。通过对似然函数的分解,得到联合信道估计与信号检测问题的因素图描述,然后利用和积算法推导出算法的迭代规则。针对迭代算法中由于信道参数引入的连续变量积分问题,提出用数值方式求解积分式的近似值,即引入粒子滤波算法,提出一种基于SP-PF的联合信道估计与信号检测的迭代算法.在上述研究的基础上,本文进一步提出一种统一结构的联合信道估计与信号检测的迭代接收机结构。该接收机将信道估计与信号检测视为两个独立的模块分别进行处理,利用基于因素图的迭代算法进行信号检测,并将获得的符号软信息传递给信道估计模块,从而实现了联合的信道估计与信号检测。在该接收机结构下,无论采用何种信道估计算法,模块间的信息交互都可以明显改善信道估计的性能,从而提高了整个接收机的信号检测性能。在对SISO系统信号检测研究的基础上,本文将上述基于因素图模型的迭代信号检测算法推广到MIMO系统的信号检测算法的研究中,提出一种MIMO系统信号检测的因素图结构,在此基础上,应用和积算法推导出信号检测的迭代算法,可以获得逼近最优检测算法的性能。为降低算法计算量,在标准和积算法基础上,提出一种MIMO系统的基于SP-PDA的快速迭代信号检测算法。更多还原

【Abstract】 As an important part of wireless communication system, wireless signal detection is faced with lots of new challenges in the next generation of mobile communication. Till now, a lot of algorithms have been proposed for the traditional single-input -single-output (SISO) system or the emerging multiple-input-multiple-output (MIMO) system. However, it is still a hot topic to obtain the optimal detection algorithm with a complexity as low as possible. Based on the existing work, this dissertation dedicates on research works on iterative detection algorithm for wireless communication by using factor graph under the information exchange theory frame.First, under the assumption of ideal channel estimation, iterative signal detection technology is studied in SISO system. By factoring of the likelihood function, the description of signal detection problem is achieved, and then iterative signal detection algorithm can be deduced by using sum-product algorithm. Furthermore, in order to reduce the complexity of sum-product algorithm, fast iterative signal detection algorithm based on SP-PDA (sum-product-probability-data-association) is proposed. Simulation results show that the iterative signal detection algorithm based on sum-product algorithm can approaching the optimal performance with computational complexity being reduced efficiently, while the fast iterative signal detection algorithm can obtain a tradeoff between performance and complexity, where computational complexity is reduced to being linear with the length of channel instead of being exponential with the length of channel.Second, optimal performance of channel equalization and signal detection relies on exact channel knowledge. Considering channel information is unknown and time-variant, joint channel estimation and signal detection based factor graph is studied. The factor graph for joint channel estimation and signal detection is set up according to the likelihood function and iterative rule is deduced by sum-product algorithm. To solve the integral of continuous variable of channel information, numerical method named as particle filter is considered to obtain its approximate result and the iterative algorithm for joint channel estimation and signal detection based on SP-PF (sum-product-particle-filter) is proposed.Furthermore, a unified iterative receiver of joint channel estimation and signal detection is designed in this dissertation. The receiver consists of two separate modules named channel estimation and signal detection, where signal detection is implemented by the iterative algorithm based on factor graph and is able to pass the soft information to help channel estimation and then the joint channel estimation and signal detection is achieved. Under the unified receiver structure, information passing between two modules can obviously improve the channel estimation performance whatever channel estimation technology is used; therefore, the signal detection performance of the whole receiver is improved.Based on the study of SISO system, the research of iterative signal detection using factor graph is extended to MIMO system. After the proposal of a new description of factor graph in MIMO system, sum-product algorithm is applied to obtain the iterative algorithm which can approximate the optimal performance. In order to reduce the complexity, PDA method is combined with the sum-product algorithm and a fast iterative signal detection algorithm based on SP-PDA is proposed.更多还原