摘要
无线移动通信的信道非常复杂，使无线通信系统的性能有很大的局限性。
发射机和接收机之间的多径问题，以及快速运动的移动终端在接收信号时产
生的多普勒效应，引起接收信号的幅度和相位都有很大的变化，就是通常所
说的衰落。也正是由于深度衰落的影响，严重的限制了通信系统的性能，需
要出现更好的调制、编码、功率控制等方法，更有效的应用于衰落信道。A.J.
Glodsmith 和 S.G..Chua 等人研究了一些新的自适应传输技术，例如自适应调
制、自适应编码、自适应功率控制，自适应天线阵列系统等。这些自适应传
输方法，都是通过瞬时监视信道条件，来调整调制水平、符号率、码率、传
输功率等级、传输天线增益或者与这些参数相关的一些方面。它们都是尽量
在不牺牲误码率性能的同时，更有效的利用功率和频谱，来实现更高的信息
传输速率。
实际中，为了实现自适应传输方法，传输过程中的信道状态信息（CSI）
必须是可知的。通过接收机一端，可以获得的过去一段时间衰落信道系数的
采样，用信道预测算法来预测未来的一段时间间隔内，衰落信道系数将会怎
样的变化，然后通过反馈信道传输给发射机，发射机根据数据传输时将会出
现的信道状态信息，来确定发射功率，调制方法，编码方法和传输天线等一
些问题，来适应当时的传输条件。这样，发射机就可以优化的进行传输，随
之而来也就提高了通信的质量。
本文首先介绍了课题的意义，发展现状和一些基础理论知识。主体部分
以及主要工作，就是对当前的信道预测算法的改进。主要分成两大部份：第
一部分就是对常用信道预测算法 ESPRIT 算法的改进部分，第二部分是自适
应子空间追踪算法的研究。改进算法有
1. 基于采样数据共轭重排的 C-ESPRIT 算法
这种方法的基本思想是：原 ESPRIT 算法仅仅利用了采样数据本身，忽
略了采样数据是复数的事实，再次利用采样数据的共轭数据，来参与自相关
函数的运算，保证计算量不变的同时提高算法的准确度和精确度。
70
吉林大学硕士学位论文
2. 基于数据预处理的 SS-ESPRIT 和 MSS-ESPRIT 算法
基本 ESPRIT 算法是目前常用的几种算法（例如最大熵算法（MEM），
长距离预测（LRP）算法）中性能最好的一种算法，它的主要缺点是进行奇
异值分解（SVD）或特征值分解时运算量较大，限制了算法的实时应用。针
对基本 ESPRIT 的这一不足，可以想办法减少奇异值或特征值分解的阶数，
提出了 SS-ESPRIT 和 MSS-ESPRIT 算法。其中 SS-ESPRIT 是应用空间平滑
技术的 ESPRIT 方法，空间平滑的基本思想是将用Rcc = CCH 方法估计的采
?
样数据自相关矩阵进行重新分块并且组合，得到阶数较少的采样数据自相关
矩阵的估计，再对其进行特征值分解。显然这种算法能够减少特征值分解时
候的运算量，但是要以牺牲运算精度为代价；而 MSS-ESPRIT 算法实质上是
利用了前两个改进方法的优点，它将采样数据共轭重排和空间平滑技术相结
合，形成了双向平滑技术，仿真实验证明，MSS-ESPRIT 算法的性能非常好，
在减少了运算量（特征值分解的阶数）的同时，也提高了预测的预测性能。
3. 子空间追踪算法
在以往的所有信道预测方法中，都是采用标准的平稳杰克模型，它由几
个复数正弦信号的和组成。重要的是假定模型中的参数幅度 An、多普勒频移
fn 和相位φn 都是固定不变的。然而实际中，这些参数都是缓慢变化着的，这
些非平稳的变化会严重的恶化信道预测性能，为了克服这一不足，采用了自
适应子空间追踪同 ESPRIT 算法相结合的方法。这种算法的优点是：首先，
子空间追踪算法解决了标准杰克衰落模型中，多普勒频移 fn 的变化带给传
统信道预测算法性能的恶化问题，这样信道预测算法能够适合更多的应用场
合；其次，它采用的是一种自适应的方法，而基本 ESPRIT 算法是批处理的
算法，运算量很大，子空间追踪算法和基本 ESPRIT 算法比起来，在准确度
和精度基本相当的情况下，运算量会减少很多，更有利于进行实时应用；再
次，还对现有的子空间追踪算法 FAST 和 FAST2 进行了改进，提出了 MFAST
和 MFAST2 算法，仿真实验和统计数字表明，它们比原有的子空间追踪算法
FAST 和 FAST2 算法不仅减少了运算量，同时也提高了预测的准确性和精确
性，加快了收敛速度，更适合于多普勒频移 fn变化严重的场合，是两种非常
好的算法。
还给出了信道预测算法在自适应传输技术中自适应调制的应用。自适应
71
摘 要
调制的基本思想就是：预测传输时刻的信道状态信息，根据信道将要发生的
条件来选择调制的具体方法。这样，总体来说在保证了误码率要求的同时，
提高了带宽效率，是一种很好的自适应传输方法。
最后总结了本次论文的全部工作内容，指出了本次论文中信道预测算法
的成绩和不足之处，对今后的信道预测算法工作重点进行了展望。
The wireless mobile communication channel is very complex. It limits the
performance of the wireless communication system a lot. The multipath problem
between the transmitter and the receiver and the fast moving terminal which cause
the Doppler affect result in the amplitude and phase of the receive signal vary
fastly. We call it “feeding”. It is the influence of the deep feeding that limit the
performance of the communication system very much. We need the better
modulation, coding and power control methods to be used in the feeding channel
efficiently. A.J. Glodsmith and S.G..Chua have investigated some new adaptive
transmitter technique, for example adaptive modulation, adaptive coding, adaptive
power control and adaptive transmitter antenna diversity and so on. These
adaptive transmitter schemes all vary the constellation size, symbol rate, coding
rate, transmitted power level weights of transmission antennas or any combination
of these parameters by instantaneously monitoring channel conditions. They are
trying to use both power and spectrum more efficiently without sacrificing the bit
error rate performance to realize the higher information transmitter rate.
To implement adaptive transmission methods in practice, the channel state
(CSI) must be available at the transmitter. With the sampling data of the receiver
samples in the past interval, we use the arithmetic of channel predictions to
predict the fading channel coefficient. Afterwards, transmit the fading channel
coefficient to transmitter in the feedback channel. The transmitter will decide the
transmission power, modulation methods, coding methods and the transmission
antennas, to fit the transmission conditions of the time. Thus, transmitter is
optimized, and the communication performance is improved at the same time.
The thesis first introduces the significance of this subject, development status
and some basic theory knowledge. The important part and mostly task of this
thesis are the modification of the current channel prediction arithmetic. It has two
parts: the first part is the improvement of basic ESPRIT arithmetic. It is a channel
prediction method in common use. The second part is the investigation of the
adaptive subspace tracing. These modified arithmetic is
1. The C-ESPRIT arithmetic based on sampling data conjugation recomposition
The basic idea is: the sampling data itself is used in the basic ESPRIT
arithmetic but the data are complex numbers is ignored. The conjugation data of
73
ABSTRACT
the sampling is reused to compute the self-correlation matrix in C-ESPRIT
arithmetic. The veracity and precision of the arithmetic are improved and the
operation quantity does not changed much at the same time.
2. The SS-ESPRIT and MSSESPRIT arithmetic based on data pretreatment
The basic ESPRIT arithmetic is the best in the current prediction arithmetic
(such as the Maximum Entropy Method (MEM) and the Long-range Prediction
method (LRP)). Its’ serious shortcoming is that the operation quantity is great
when the singular value decomposition (SVD) or eigenvalue decomposition is
computed. It limits the real time application. To overcome the shortcoming of the
arithmetic, the dimensions of singular value decomposition or eigenvalue
decomposition must be reduced. SS-ESPRIT and MSS-ESPRIT arithmetic are
given. The SS-ESPRIT is formed by the basic ESPRIT method combined with the
space smooth techniques. The basic idea of the space smooth techniques is that
the self-correlation matrix is disparted and assembled again, which is estimated by
the Rcc = CCH method. Then the eigenvalue decomposition is computed. It is
?
apparent that the operation quantity is reduced by reducing the dimensions of
eigenvalue decomposition but the prediction precision is decreased. Well the
virtues of the two arithmetics are utilized by the MSS-ESPRIT arithmetic
substantially. The sampling data conjugation recomposition and space smoothing
techniques are com
引文
[1] A.J. Glodsmith and S.G..Chua, “Adaptive modulation for fading channels”,
IEEE Trans. Commun. May 1998, Vol.46, pp.595-601
[2] S.Otsuki, S.Sampei, and N.Mornaga, “Squre-QAM adaptive modulation
/TDMA/TDD system using modulation level estimation with Walsh function”,
Electron. Lett., Feb 1995, Vol.31, pp.169-171
[3] W.T.Webb and R.Steele, ”Variable rate QAM for mobile radio”, IEEE Trans.
Commun. July 1995, Vol.43, pp.2223-2230
[4] A.J. Glodsmith and S.G..Chua, “Variable –rate variable-power MQAM for
fading channels”. IEEE Trans. Commun. Oct 1997, Vol.45, pp.1218-1230
[5] B.Vucetic,”An adaptive coding scheme for time-varying channels”, IEEE
Trans. Commun. May 1991, Vol.39, pp.653-663
[6] R.Chen.K.C.Chua, B.T.Tan, and C.S.Ng, ”adaptive error coding using channel
prediction,” in Proc. IEEE Int. Symp. Personal, Indoor and Mobile Radio
Commun. PLMRC’96, 1996, Vol.1996.pp.359-363
[7] B.Narendran, J.Sienicki, S.Yajinik, and P.Afrawal, “Evaluation of an adaptive
power and error control algorithm for wireless syetems”,in Proc IEEE Int.
Commun. Conf., ICC’1997.1997, Vol 1, pp.149-355
[8] P.Agrawal, B.Narendran,J,sieniki, and S.Yajnik, “an adaptive power control
and coding scheme for mobile radio system”, in Proc. IEEE Int. Conf. Personal
Wireless Commun, ICPWC’96, 1997, pp.283-288
[9] K.Rohani, M.Harrson, and K.Kuchi, ”A comparison of base station transmit
diversity methods for third generation cellular standards”, in Proc. IEEE Veh.
Technol. Conf. VCT’99, 1999, Vol.1,pp.351-355
[10] T.K.Yo,”Maximum raion transmission”, IEEE Trans. Commun.Oct. 1999,
Vol.47, pp.1458-1461
[11] J.Lin. J.G.Proakis, Fling, and H.Lev-Ari, ”Optimal Tracking for Time Varying
Channels: A Frequency Domain Approach for Known New algorithms”,
IEEE Transactions on Selected Areas in Communications, Jan. 1995,
13(1):141-154
[12] R.Haeb and H.Mayr, “A systematic Approach to Carrier Recovery and
Detection of Digitally Phase Modulated Signals on Fading Channels”, EEE
Transactions on Communication, July 1989, 37(7):748-754
[13] T.Eyceoz,A.Duel-Hallen,and H.Hallen,“Prediction of Fast Fading Parameters
65
参考文献
by Resolving the Interference Pattern”,Proceeding of the 31st ASLOMAR
Conference on Signals Systems,and computers, 1998. pp.167-171,November
2-5
[14] A.Duel-Hallen,Shengquan Hu,and H.Hallen, “ Long-range Prediction of
Fading Signals”, Processing Magazine, May 2000. pp.62-75
[15] J φ rgen Bach Andersen,Jesper Jenson,S φ ren Holdt Jensen,and Frank
Frederiksen, “ Prediction of Future Fading Based on Past
Measurements”,IEEE VTC’99, 1999, pp.151-155
[16] J.K.Kwang and J.H.Winters, ”Sinusoidal modeling and perdiction of fast
fading processes”, in Proc.IEEE GLOBECOM’98,pp. 892-897.
[17] X.M.Gao, M.A.Tanskanen, and S.J.Ovaska, “Comparison of linear and neural
network based power prediction schemes for mobile DS/CDMA systems,” in
Proc. IEEE Veb. Technol. Conf.,VTC’96, May 1996, pp.61-65
[18] Y.S.Zhang and D.B.Li, ”Volterra adaptive prediction of multipath fading
channel,” Electron. Lett., Apr.1997, Vol.33, No.9, pp.754-755.
[19] C.Spillard and G.J.R.Povey,”Application of the Prony Algorithm to a
predictive RAKE receiver,”in Proc. IEEE 4th Int. Symp. Spread Spectrum
Tecb, and Applicaions, 1996, Vol.3, pp.1039-1042
[20] Edward, C. Real, Donald W.Tufts, ”Two Algorithms for Fast Approximate
Subspace Tracking”, IEEE Trans. signal Processing. July 1999, Vol.47, No.7,
pp. 1936-1945
[21] 郭梯云、杨家玮、李建东，“数字移动通信”，人民邮电出版社
[22] John G.Proakis, 张力军，张宗橙，郑宝玉等译，“数字通信”，电子工业
出版社
[23] Daniel, J.Rabideau, “Fast, Rank Adaptive Subspace Tracking and
applications”, IEEE Trans. signal Processing, vol. 44, No.9, pp. 2229-2244,
Sep.1996.
[24] 薛强、杜志敏、吴伟陵，“一种基于一阶扰动自适应特征值分解的多用户
检测算法”，电子学报，vol.15, No.4, Nov, 2000, pp.413-416
[25] Daniel, J.Rabideau, “Fast, Rank Adaptive Subspace Tracking and
applications”, IEEE Trans. signal Processing, Sep.1996. vol. 44, No.9,
pp. 2229-2244
[26] K.J.Ray Liu, Dianne P.O’Leary, G.W.Stewart and Yuan-Jye J.Wu, “URV
ESPRIT for Tracking Time-Varying Signals”, IEEE Trans. signal Processing,
Dec.1994vol. 44, No.12, pp. 3441-3448.
[27] T.Eyceoz,Shengquan Hu,and A.Duel-Hallen,H.Hallen, “Adaptive Prediction,
Tracking and Power Adjustment for Frequency Non-Selective Fast
66
吉林大学硕士学位论文
Channels”,Proceedings of the International Conference on Communications
(ICC’99), pp.1-5
[28] Shengquan Hu, A.Duel-Hallen,and H.Hallen,“Long Range Prediction Makes
Adaptive Modulation Feasible for Realistic Mobile Radio
Channels”,Proceeding of the 34rd Annual Conference on Information
Sciences and Systems(ciss’00), pp.1-6
[29] Torbjorn Ekman,Gernot Kubin,“Nonlinear Prediction of Mobile Radio
Channels: Measurements and MARS Model Designs ” , IEEE, 1999,
pp.2667-2670.
[30] Buogang Hu,Raymond G.Gosine, “ A New Eigenstructure Method for
Sinusoidal Signal Retrieval in White Noise Estimation and Pattern
Recognition”,IEEE Transactions on Processing.Vol.45.No.12,December 1997,
pp.3073-3083
[31] 美 S.M.凯依，“现代谱估计原理与应用”，科学出版社
[32] 潘示先，“谱估计和自适应滤波”，北京航空航天大学出版社
[33] 张贤达，“现代信号处理”，清华大学出版社
[34] 张贤达，“信号处理中的线性代数”，科学出版社
[35] 陈尚勤、李晓峰，“快速自适应信息处理”，人民邮电出版社
[36] 美 Vinay K.Ingle Hohn G.Proakis, 陈怀琛、王朝英、高西全译，“数字信
号处理及其 MATLAB 实现”，电子工业出版社
[37] 蒋同泽，“现代移动通信系统”。电子工业出版社
[38] 姚庆栋、梁慧君、戴文琪。“数字无线传输”，浙江大学出版社
[39] 肖维民、彭应宁，“两种快速 ESPRIT 算法”,电子学报，July , 1997. Vol,25.
No.10, pp.102-104
[40] 俞胜，陈光禹，“一种检测噪声中正弦信号的 SVD 方法”，电子学报， vol.28,
No.6, June 2000, pp.108-110
[41] 冯晋利、武洁、王绍卿，“快速高分辨率估计算法”，现代引信，1998 年
第一期，pp.53-57
[42] 何子述、黄振兴、向敬成，“基于共轭数据重排的修正 ESPRIT 信号 DOA
估计算法”，信号处理，vol.17, No.1, Feb, 2001, pp.76-79
[43] 叶中付，“一种快速 TLS-ESPRIT 算法”，信号处理， vol.14, No.2, June, 1998,
pp.146-150
[44] 冯晋利、黄建国，“均匀线阵列下的快速 TLS-ESPRIT 算法”，信号处理，
vol.15, No1, Dec, 1999, pp.58-62
[45] 薛强、杜志敏、吴伟陵，“一种基于一阶扰动自适应特征值分解的多用户
检测算法”，电子学报，vol.15, No.4, Nov, 2000, pp.413-416
[46] 魏红、何佩琨、毛二可，“利用 ESPRIT 方法实现信号频率/相位的联合估
67
参考文献
计”，北京理工大学学报，vol.19, No.3, June, 1999, pp.348-351
[47] 庞潼川、何佩琨、高梅国、毛二可，“一种快速的测频算法”，系统工程
与电子技术，vol.2, 第 23 卷，第十期
[48] 周云钟，陈天麒，“一种自适应 ESPRIT 算法”，系统工程与电子技术，
vol.2, 第 23 卷，第九期
[49] 斯德谊，乐强，沈士团，“利用 Toeplitz 特性改善来波方向估计性能”，
北京航天航空大学报”vol.24, No.3, June,1998
[50] 陈辉，王永良，“空间谱估计算法结构及仿真分析”，系统工程与电子技
术，vol.23, No.8, 2001
