【作者】 赵丹；

【导师】 王建新； 于全；

【作者基本信息】 国防科学技术大学 ， 信息与通信工程， 2008， 博士

【摘要】 OFDM(Orthogonal Frequency Division Multiplexing)技术作为一种有效对抗多径衰落信道的调制技术,在军事通信领域有着很好的应用前景。跳频OFDM(FH-OFDM)系统通过射频频率的不断切换来获得抗干扰的能力,这使得它同时具有了连续和突发两种通信模式的特征,对载波频偏估计(Carrier FrequencyOffset,CFO)和信道估计技术提出了更高的要求,具体表现为在极大地提高系统的传输效率的同时保证CFO和信道估计算法的收敛速度和估计性能。同时暂态过程也成为影响系统性能的重要因素之一,必须采取应对措施。为了提高系统的传输效率,本文在给出了两种前导字较短时的CFO与信道捕获算法的基础上,重点研究了两种无需内插导频的同步与信道联合跟踪算法,并对FH-OFDM系统中相位畸变抑制技术进行了深入研究,获得以下研究成果。1.以CFO和信道估计的CRB(Cramer-Rao Bound)作为统一衡量标准,以最大似然(Maximum Likelihood,ML)估计准则为统一的估计准则,系统地分析了CFO和信道对彼此的估计性能的影响以及联合估计时方差可以达到的下限。得出结论:信道完全未知条件下的联合估计中,CFO的JCRB比信道统计特性已知条件下的独立估计的CRB更低,而信道的JCRB可逼近理想条件下信道估计的CRB。2.提出了一种面向判决(Directed Decision,DD)的同步和信道联合估计算法,得到了一定误码率条件下CFO与信道估计的均方差的表达式。为了提高算法的性能,针对不同的信道环境提出了开环和闭环两种不同结构的算法,通过理论分析和仿真验证,给出了在连续传输和突发传输两种模式下闭环结构算法中环路因子的取值准则。考虑到面向判决法所存在的误码扩散以及严重衰落信道下存在误码平台效应,将该算法与信道优选算法相结合。提出以归一化的矢量信号误差作为信道优选的度量值,与传统信道选择算法相比具有更强的鲁棒性。3.提出了两种基于期望最大(EM)算法的CFO和信道联合估计算法,根据处理域的不同分别称之为时域和频域EM联合估计算法。从提高性能和降低复杂度的角度给出了它们的改进算法——基于Bayesian准则的EM算法和逐子信道处理的频域EM估计算法。根据数据重构的方式的不同,将它们与APP(A PosterioriProbabilities)检测器和SISO(Soft-in Soft-Out)译码器结合使用,实现了CFO、信道和数据的联合检测,克服了面向判决法的误码扩散问题,大大提高了系统性能。对两种EM算法的的估计性能、复杂程度和适应的环境进行了分析和对比,频域EM算法具有更低的复杂度,并且更适合于严重衰落信道和高阶调制系统。4.针对时变信道,为CFO与信道建立了新的状态转移模型,在此基础上利用Kalman平滑算法实现了CFO与信道的联合跟踪,该算法可以归纳为EM-Kalman算法。与传统EM-Kalman算法相比,适当增加本算法的迭代次数,可以将系统性能损失控制在0.5dB以内的同时提高了10%以上的带宽利用率。5.FH-OFDM系统中的相位畸变的来源由相位抖动和杂散信号组成,确立了与之对应的统计模型,分析了相位畸变对系统性能的影响,得到了MPSK调制方式下统一的误码率表达式。在此基础上,针对相位抖动和杂散信号统计特性的不同,提出了两种相位畸变抑制算法,它们分别基于LS(Least Square)准则和MMSE(Minimum Mean Square Error)准则实现,直接对均衡器进行估计,在较少的导频符号的辅助下,以较低的复杂度实现了对相位畸变的抑制。更多还原

【Abstract】 Orthogonal Frequency Division Multiplexing (OFDM) is an effective technique for high speed transmission which is commendable to be applied in the region of military communication. Frequency Hopping OFDM (FH-OFDM) system improves its anti-jamming ability by switching frequencies in succession, which makes it bearing with the features of both continuation and burst communication mode. It comes up with higher requirement on the carrier frequency offset (CFO) estimation and channel estimation, which requires to guarantee the estimated performance and to improve the transmission efficiency as much as possible at the same time. Moreover the transient state during the communication would become a significant factor impacting the system performance.In conventional OFDM systems, the pilots are interpolated for the tracking of residual CFO and channel, which expend a lot of system resources. In order to improve the utilization ratio of bandwidth, we propose two new joint CFO and channel estimation algorithms which need no pilots. The phase distortion suppression technique for RF FHOFDM systems is also studies. The main research findings are shown as follows.1. By choosing Cramer-Rao Bound (CRB) as the unified reference point and Maximum Likelihood (ML) criteria as the unified estimation criteria for CFO and channel estimation, the impaction between CFO and channel estimation and the performance upper limit of joint estimation are analyzed systemically. The conclusion is drawn that the JCRB (Joint CRB) is beeter than the CRB of dependent estimation where the statistical characteristicr is known for CFO and is closed to the CRB in ideal conditions for the channel.2. Propose a joint Detected-Directed (DD) CFO and channel estimation technique, and obtain the expressions of Mean Square Error (MSE) of the CFO and channel estimation results with a certain Symbol Error Rate (SER) by theoretical derivation. In order to improve the estimation performance furtherly, open loop and closed loop schemes are suggested for different SNR ( Signal Noise Ratio) and channel environment. The rules that how to set the values of the loop factors are also suggested for both continuous transmission and burst transmission modes. In consideration of the problems of error spreading and error platform phenomenon, this algorithm should be assisted with the channel preference algorithm. We use the normalized Vector Signal Errors (NVSE) as the metric for the algorithm which is more robust than conventional ones.3 . Propose two CFO and channel joint estimation algorithm based on Expectation-Maximization (EM) algorithm, which are named as time domain EM (TEM) and frequency domain (FEM) algorithms according to the disposal domain respectively. The refinement algorithms to improve the performance or decrease the complexity are also proposed. The former is constructed on Bayesian criteria for both two schemes and the latter is disposed subchannel by subchannel for FEM. According to the methods of data reconstruction, the algorithm can be combined with the A Posteriori Probabilities (APP) detector or Soft-In Soft-Out (SISO) decoder to implement the joint detection of CFO, channel and data, which can improve the estimated performance greatly. The estimation performance, complexity and application settings of the two EM algorithms are analyzed and compared. The algorithm disposed in the frequency domain has lower complexity and better performance for deep fade channels and high order modulation systems. They all conquer the shortcoming of error spreading.4. Build the new state transition model for CFO and channel parameters in time varying channels. The joint estimation algorithm based on EM criteria is proposed which can be realized with EM-Kalman algorithm and tracking the time varying channel effectively. As compared with the conventional EM-Kalman algorithm based on pilots, it can improve the utilization ratio of bandwidth more than 10% and make the performance loss less than 0.5dB at the same time.5. The phase distortion of RF FH -OFDM system is made up of phase jitter and spurs. We build the corresponding statistical model for them and analyze the impact one system performance. The unified Bit Error Rate (BER) expression is made for MPSK modulation systems. Based on the analysis above, the corresponding phase jitter suppression and spurs suppression algorithm are proposed which are based on Least Square (LS) and Minimum Mean Square Error (MMSE) criteria respectively. We estimate the equalizer directly and reduce the effects of phase distortion greatly with a few pilots.更多还原