基于神经网络的多用户检测技术研究
|
摘要
CDMA系统多用户检测技术是第三代移动通信系统的关键技术之一。在CDMA通信系统中,指定给各用户的特征序列总是存在一定的相关性,这就是多址干扰产生的根源。多用户检测技术在传统检测技术的基础上,充分利用所有用户的信息来对接收信号进行联合检测,从而具有很好的多址干扰抑制性能,同时也解决了远近效应问题,降低了系统对功率控制精度的要求,显著提高了整个系统容量。
本论文在对CDMA通信系统、多用户检测模型和多用户检测性能测度等问题进行分析的基础上,介绍了最佳多用户检测原理和次佳多用户检测方法及分类。通过对Hopfield神经网络多用户检测器的分析,将该检测器求解最优多用户检测目标函数的最小值问题转化为求解Hopfield神经网络能量函数最小值的问题。针对Hopfield神经网络易陷入能量函数局部最小点的问题,本文给出了简化检测器与软判决反馈随机Hopfield神经网络相结合的一种混合Hopfield神经网络多用户检测模型。混合Hopfield神经网络多用户检测器中的简化检测器能够给Hopfield神经网络提供较好的初始值,从而避免了由于初始值的随机选取使能量函数收敛到局部最小点的情况,得到了全局意义下的满意解,提高了检测性能。仿真结果表明,混合Hopfield神经网络多用户检测有更好的抗多址干扰和抗远近效应性能,并且具有良好的收敛性。
另外,本论文还研究了混沌神经网络模型,详细的分析了Chen和Aihara提出的混沌神经网络及暂态混沌神经网络模型结构和动力学特性。暂态混沌神经网络具有比Hopfield网络更为丰富的动力学特性,可有效提高网络全局最小值的搜索能力。
在暂态混沌神经网络的基础上用模拟退火策略对暂态混沌神经网络进行了优化,在保证准确性的基础上,加快收敛速度。对优化后的暂态混沌神经网络在多用户检测技术中的应用进行了研究分析,实验仿真结果证明该神经网络检测器具有较好的检测性能,具有一定的理论意义和实用价值。
The CDMA multi-user detection is one of the main technologies of the third generation communication system. In CDMA communication system, there always have some relativity in the sequence assigned to the users, which is the root of multiple access interference. Multi-user detection (MUD) makes joint detection on the received signals by making full use of the information of all users' based on tradition detection technology. So it has good anti-interference capability, can solve far and near effect problem, reduce demand for power control precision, notability advance capability of whole system.
This paper introduces the principle of the optimal MUD and some classical suboptimum MUD based on the research on the model of CDMA communication system and the principle of MUD. It introduces in detail the detection based on Hopfield neural network (HNN), which translates the problem of minimizing the objective function of the optimal MUD into resolving the minimization of energy function for HNN and has the disadvantage of easily getting into the local minimization of energy function. Aim at this disadvantage, this paper gives a model of hybrid MUD based on reduced detection and the stochastic HNN with soft decision feedback. The reduced detection of hybrid detection provides a better initial value to HNN, then HNN will not get into the local minimization of energy function and get the global optimal solution to improve detection performance. The simulation results indicate that hybrid MUD has better performance in anti-multiple access interference and far and near resistance, and has a good convergence.
This paper researches the structure and elements of chaos neural network. It analyzes the structure and dynamics of chaos neural network and transient chaos neural network proposed by Chen and Aihara. The transient chaos neural network has more abundant dynamics characteristic neural network than the Hopfield neural network. It can advance the network's searching ability of global minimum efficiencily.
This paper optimizes transient chaos neural network with simulated annealing mechanics. It can accelerate the search speed and guarantee the assurance of the veracity of the optimal arithmetic. The improved transient chaos neural network is researched and applied to the MUD. Experiment proves that this MUD has good performance and there has some theoretic meaning and practicality value in it.
本论文在对CDMA通信系统、多用户检测模型和多用户检测性能测度等问题进行分析的基础上,介绍了最佳多用户检测原理和次佳多用户检测方法及分类。通过对Hopfield神经网络多用户检测器的分析,将该检测器求解最优多用户检测目标函数的最小值问题转化为求解Hopfield神经网络能量函数最小值的问题。针对Hopfield神经网络易陷入能量函数局部最小点的问题,本文给出了简化检测器与软判决反馈随机Hopfield神经网络相结合的一种混合Hopfield神经网络多用户检测模型。混合Hopfield神经网络多用户检测器中的简化检测器能够给Hopfield神经网络提供较好的初始值,从而避免了由于初始值的随机选取使能量函数收敛到局部最小点的情况,得到了全局意义下的满意解,提高了检测性能。仿真结果表明,混合Hopfield神经网络多用户检测有更好的抗多址干扰和抗远近效应性能,并且具有良好的收敛性。
另外,本论文还研究了混沌神经网络模型,详细的分析了Chen和Aihara提出的混沌神经网络及暂态混沌神经网络模型结构和动力学特性。暂态混沌神经网络具有比Hopfield网络更为丰富的动力学特性,可有效提高网络全局最小值的搜索能力。
在暂态混沌神经网络的基础上用模拟退火策略对暂态混沌神经网络进行了优化,在保证准确性的基础上,加快收敛速度。对优化后的暂态混沌神经网络在多用户检测技术中的应用进行了研究分析,实验仿真结果证明该神经网络检测器具有较好的检测性能,具有一定的理论意义和实用价值。
The CDMA multi-user detection is one of the main technologies of the third generation communication system. In CDMA communication system, there always have some relativity in the sequence assigned to the users, which is the root of multiple access interference. Multi-user detection (MUD) makes joint detection on the received signals by making full use of the information of all users' based on tradition detection technology. So it has good anti-interference capability, can solve far and near effect problem, reduce demand for power control precision, notability advance capability of whole system.
This paper introduces the principle of the optimal MUD and some classical suboptimum MUD based on the research on the model of CDMA communication system and the principle of MUD. It introduces in detail the detection based on Hopfield neural network (HNN), which translates the problem of minimizing the objective function of the optimal MUD into resolving the minimization of energy function for HNN and has the disadvantage of easily getting into the local minimization of energy function. Aim at this disadvantage, this paper gives a model of hybrid MUD based on reduced detection and the stochastic HNN with soft decision feedback. The reduced detection of hybrid detection provides a better initial value to HNN, then HNN will not get into the local minimization of energy function and get the global optimal solution to improve detection performance. The simulation results indicate that hybrid MUD has better performance in anti-multiple access interference and far and near resistance, and has a good convergence.
This paper researches the structure and elements of chaos neural network. It analyzes the structure and dynamics of chaos neural network and transient chaos neural network proposed by Chen and Aihara. The transient chaos neural network has more abundant dynamics characteristic neural network than the Hopfield neural network. It can advance the network's searching ability of global minimum efficiencily.
This paper optimizes transient chaos neural network with simulated annealing mechanics. It can accelerate the search speed and guarantee the assurance of the veracity of the optimal arithmetic. The improved transient chaos neural network is researched and applied to the MUD. Experiment proves that this MUD has good performance and there has some theoretic meaning and practicality value in it.
引文
[1]袁超伟,陈德荣,冯志勇.CDMA蜂窝移动通信.北京:北京邮电大学出版社,2003
[2]曹达仲,候春萍.移动通信原理、系统技术.北京:清华大学出版社,2004:6-8页
[3]Hamouda.A fast adaptive algorithm for MMSE receivers in DS-CDMA systems.IEEES.P.Letters,2004,11(4):86-89P
[4]K.S.Schneider.Optimum detection of code division multiplexed signals.IEEETrans.Aerosp.Electron.Syst.,Vol.AES-15,No.1,Jan.1979:181-185P
[5]S.Verdu.Multi-user Detection in Advances in Statistical Signal Processing,Vol.2,NewYork:JA.I Press,1993:369-409P
[6]张贤达,保铮.通信信号处理.北京:国防工业出版社,2000:420-433页
[7]刘向东,顾学迈.第三代移动通信系统中非线性多用户检测技术.华北航天工业学院学报.2004.3(1):21-23页
[8]Aazhang B.,B.P.Paris and G..Orsak.Neural Network for multiuser detection in code-division multiple access systems.IEEE Trans.Commun.,Vol.40,No.7,1992:1212-1222P
[9]R.Tanner and D.G.M.Cruickshank.Non-linear Volterra Filer Receiver for DS-CDMA.In Proceedings 4~(th) International Conference on Mathematics in Signal Processing,IMA,University of Warwick,UK,IEEE,December 1996
[10]U.Mitra and H.V.Poor.Neural Network Techniques for Adaptive Multiuser Demodulation.IEEE Transactions on Sel.Areas.in Commun.,Vol.12,Dec.1994:1460-1470P
[11]G.I.Kechriotis and E.S.Manolakos.Hopfield neural-network implementation of the optimal CDMA multiuser detector.IEEE Trans. Neural Networks,vol.7,1996:131-141P
[12]G.Jeney,J.kevendovszky.Stochastic hopfield network for multi-user detection.European Conf.Wireless Techn.Paris,2000:147-150P
[13]Aihara K,Takabe T,and Toyoda M.Chaotic neural networks,Phys.Lett.A,1990,144(6/7):333-340P
[14]B.Wang,J.Nie,and Z.He.A transiently chaotic neural network implementation of the cdma multiuser detector.IEEE Trans.Neural Networks,vol.10,1999:1257-1259P
[15]W.Zhong and S.X.Cheng,Multiuser detection using time-varying scaling-parameter transiently chaotic neural networks.ELECTRONICS LETERS 10~(th) Vol.35 No12,June 1999:987-989P
[16]戴一昊,蒋铃鸽,何晨.自适应暂态混沌神经网络在CDMA多用户检测器中的应用.上海交通大学学报.第38卷第5期.2004:697-700页
[17]Sumit Bhardwaj and Jayadeva.Sequential chaotic annealing neural network for CDMA multiuser detection.Proceedings of the 9~(th)International Conference on Neural Information Processing(ICONIP'OZ).Vol.5:2176-2180P
[18]X.Wang and H.V.Poor.Iterative(Turbo) soft interference cancellation and decoding for code CDMA.IEEE Trans.Commun.Vol.47,No.7,July 1999:1046-1061P
[19]Z.G.Yang and X.D.Wang.Blind turbo multiuser detection for long-code multipath CDMA.IEEE Trans.On Commun.,Vol.50,No.1 JAN.2002:112-125P
[20]Honig M,Madhow U,Verdu S.Blind adaptive multiuser detection.IEEE Trans.Inform.Theory,1995.41:944-960P
[21]Wang X.and Poor H.V.Blind multiuser detection:A subspace approach.IEEE Trans.IT.,Vol.44,1998:677-691P
[22]Zhang X D,Wei W.Blind adaptive multiuser detection based on Kalman filtering.IEEETrans.Signal Processing,1994(42):2854-2858P
[23]Anton-Haro C,Fonollosa J A R.Probabilistic algorithms for blind adaptive multiuser detection.IEEE Trans.Signal Processin,1995
[24]Woodward G and Vucetic B.Adaptive detection for DS-CDMA.Proceeding of the IEEE.July 1998,86(7):1413-1434P
[25]Madhow U.MMSE interference suppression for timing acquisition and demodulation in direct-sequence CDMA systems.IEEE Trans.Commun.,1998(46):1065-1075P
[26]Verdu S.Minimum probability of error for asynchronous gaussian multi-access channels.IEEE Trans.Inform.Theory,1986,32(1):85-96P
[27]Varansi M.K,Aazhang B.Multistage detection in asynchronous code-division communications.IEEE Trans.On Commun,1990,38(4):509-519P
[28]Simon Haykin著,叶世伟,史忠植译.神经网络原理.北京:机械工业出版社,2004:497-503页
[29]李士勇.模糊控制.神经控制和智能控制论.哈尔滨:哈尔滨工业大学出版社,1998:15-118页
[30]靳蕃.神经计算智能基础原理方法.西南交通大学出版社,2000
[31]阎平凡,张长睡.人工神经网络与模拟进化算法.北京:清华大学出版社,2000
[32]Martin T.Hagan,Howard B.Demuth著.戴葵等泽.神经网络设计.北京:机械工业出版社,2002:399-409页
[33]N.Mana,R.B Ayed,R.Bouallegue.A Software Simulation Tested for Multi-user Detection Based on Modified Hopfield Network.First International Symposium on Control,Communications and Signal Processing,2004:335-338P
[34]王永刚,焦李成.基于随机Hopfield神经网络的最优多用尸俭测.电子学报.2004,32(10):1630-1634页
[35]韩靖,郑宝玉.基于Hopfield神经网络的DS-CDMA多用户检测.南京邮电学报,1998,18(2):1-4页
[36]B.Porat,B.Friedlander.Parametric Techniques for Adaptive Detection of Gaussian Signals.IEEE Transactions on Signal Processing.1994,32(8): 780-790P
[37]W G Teich,M Seidl.Code division multiple access communication:Multi-user detection based on a recurrent structure.IEEE Trans.Veh.Teclmol.,1996(7):979-984P
[38]Wang Yong gang,Li yah.ANovel Detector Using the Stochastic Hopfield Network with Soft Decision Feedback.Proceeding of IEEE TENCON'02.2002:1024-1027P
[39]George Kechriotis and Elias S.Manolakos.A Hybrid Digital Signal Processing-Neural Network CDMA Multiuser Detection Scheme.IEEE Transactions on Circuits and Systems-Ⅱ:Analog and Digital Signal Processing,1996,vol.43:96-105P
[40]焦李成.神经网络系统理论.西安:西安电子科技大学出版社,1990
[41]Wang L,Smith KA.On chaotic simulated annealing.IEEE Trans.Neural Net.1998(9):716-718P
[42]Chen,L.,Aihara,K.Chaotic simulated annealing by a neural network model with transient chaos.Neural networks.1995,8(6):915-930P
[43]李薪宇,吕炳朝.暂态混沌神经网络中的模拟退火策略优化.计算机应用.2005,25(10):2410-2412页
[44]谭营,王保云,何振亚.一种具有暂态混沌和时变增益的神经网络及其在优化计算中的应用.电子学报.1998,25(7):818-822页
[45]张强,马润年,许进.一种混沌神经网络模型及其在优化中的应用.系统工程与电子术.2002,24(2):48-50页
[46]谢传泉,何晨.混沌神经网络模型中的模拟退火策略.上海交通大学学报.2003,37(3):36-39页
[2]曹达仲,候春萍.移动通信原理、系统技术.北京:清华大学出版社,2004:6-8页
[3]Hamouda.A fast adaptive algorithm for MMSE receivers in DS-CDMA systems.IEEES.P.Letters,2004,11(4):86-89P
[4]K.S.Schneider.Optimum detection of code division multiplexed signals.IEEETrans.Aerosp.Electron.Syst.,Vol.AES-15,No.1,Jan.1979:181-185P
[5]S.Verdu.Multi-user Detection in Advances in Statistical Signal Processing,Vol.2,NewYork:JA.I Press,1993:369-409P
[6]张贤达,保铮.通信信号处理.北京:国防工业出版社,2000:420-433页
[7]刘向东,顾学迈.第三代移动通信系统中非线性多用户检测技术.华北航天工业学院学报.2004.3(1):21-23页
[8]Aazhang B.,B.P.Paris and G..Orsak.Neural Network for multiuser detection in code-division multiple access systems.IEEE Trans.Commun.,Vol.40,No.7,1992:1212-1222P
[9]R.Tanner and D.G.M.Cruickshank.Non-linear Volterra Filer Receiver for DS-CDMA.In Proceedings 4~(th) International Conference on Mathematics in Signal Processing,IMA,University of Warwick,UK,IEEE,December 1996
[10]U.Mitra and H.V.Poor.Neural Network Techniques for Adaptive Multiuser Demodulation.IEEE Transactions on Sel.Areas.in Commun.,Vol.12,Dec.1994:1460-1470P
[11]G.I.Kechriotis and E.S.Manolakos.Hopfield neural-network implementation of the optimal CDMA multiuser detector.IEEE Trans. Neural Networks,vol.7,1996:131-141P
[12]G.Jeney,J.kevendovszky.Stochastic hopfield network for multi-user detection.European Conf.Wireless Techn.Paris,2000:147-150P
[13]Aihara K,Takabe T,and Toyoda M.Chaotic neural networks,Phys.Lett.A,1990,144(6/7):333-340P
[14]B.Wang,J.Nie,and Z.He.A transiently chaotic neural network implementation of the cdma multiuser detector.IEEE Trans.Neural Networks,vol.10,1999:1257-1259P
[15]W.Zhong and S.X.Cheng,Multiuser detection using time-varying scaling-parameter transiently chaotic neural networks.ELECTRONICS LETERS 10~(th) Vol.35 No12,June 1999:987-989P
[16]戴一昊,蒋铃鸽,何晨.自适应暂态混沌神经网络在CDMA多用户检测器中的应用.上海交通大学学报.第38卷第5期.2004:697-700页
[17]Sumit Bhardwaj and Jayadeva.Sequential chaotic annealing neural network for CDMA multiuser detection.Proceedings of the 9~(th)International Conference on Neural Information Processing(ICONIP'OZ).Vol.5:2176-2180P
[18]X.Wang and H.V.Poor.Iterative(Turbo) soft interference cancellation and decoding for code CDMA.IEEE Trans.Commun.Vol.47,No.7,July 1999:1046-1061P
[19]Z.G.Yang and X.D.Wang.Blind turbo multiuser detection for long-code multipath CDMA.IEEE Trans.On Commun.,Vol.50,No.1 JAN.2002:112-125P
[20]Honig M,Madhow U,Verdu S.Blind adaptive multiuser detection.IEEE Trans.Inform.Theory,1995.41:944-960P
[21]Wang X.and Poor H.V.Blind multiuser detection:A subspace approach.IEEE Trans.IT.,Vol.44,1998:677-691P
[22]Zhang X D,Wei W.Blind adaptive multiuser detection based on Kalman filtering.IEEETrans.Signal Processing,1994(42):2854-2858P
[23]Anton-Haro C,Fonollosa J A R.Probabilistic algorithms for blind adaptive multiuser detection.IEEE Trans.Signal Processin,1995
[24]Woodward G and Vucetic B.Adaptive detection for DS-CDMA.Proceeding of the IEEE.July 1998,86(7):1413-1434P
[25]Madhow U.MMSE interference suppression for timing acquisition and demodulation in direct-sequence CDMA systems.IEEE Trans.Commun.,1998(46):1065-1075P
[26]Verdu S.Minimum probability of error for asynchronous gaussian multi-access channels.IEEE Trans.Inform.Theory,1986,32(1):85-96P
[27]Varansi M.K,Aazhang B.Multistage detection in asynchronous code-division communications.IEEE Trans.On Commun,1990,38(4):509-519P
[28]Simon Haykin著,叶世伟,史忠植译.神经网络原理.北京:机械工业出版社,2004:497-503页
[29]李士勇.模糊控制.神经控制和智能控制论.哈尔滨:哈尔滨工业大学出版社,1998:15-118页
[30]靳蕃.神经计算智能基础原理方法.西南交通大学出版社,2000
[31]阎平凡,张长睡.人工神经网络与模拟进化算法.北京:清华大学出版社,2000
[32]Martin T.Hagan,Howard B.Demuth著.戴葵等泽.神经网络设计.北京:机械工业出版社,2002:399-409页
[33]N.Mana,R.B Ayed,R.Bouallegue.A Software Simulation Tested for Multi-user Detection Based on Modified Hopfield Network.First International Symposium on Control,Communications and Signal Processing,2004:335-338P
[34]王永刚,焦李成.基于随机Hopfield神经网络的最优多用尸俭测.电子学报.2004,32(10):1630-1634页
[35]韩靖,郑宝玉.基于Hopfield神经网络的DS-CDMA多用户检测.南京邮电学报,1998,18(2):1-4页
[36]B.Porat,B.Friedlander.Parametric Techniques for Adaptive Detection of Gaussian Signals.IEEE Transactions on Signal Processing.1994,32(8): 780-790P
[37]W G Teich,M Seidl.Code division multiple access communication:Multi-user detection based on a recurrent structure.IEEE Trans.Veh.Teclmol.,1996(7):979-984P
[38]Wang Yong gang,Li yah.ANovel Detector Using the Stochastic Hopfield Network with Soft Decision Feedback.Proceeding of IEEE TENCON'02.2002:1024-1027P
[39]George Kechriotis and Elias S.Manolakos.A Hybrid Digital Signal Processing-Neural Network CDMA Multiuser Detection Scheme.IEEE Transactions on Circuits and Systems-Ⅱ:Analog and Digital Signal Processing,1996,vol.43:96-105P
[40]焦李成.神经网络系统理论.西安:西安电子科技大学出版社,1990
[41]Wang L,Smith KA.On chaotic simulated annealing.IEEE Trans.Neural Net.1998(9):716-718P
[42]Chen,L.,Aihara,K.Chaotic simulated annealing by a neural network model with transient chaos.Neural networks.1995,8(6):915-930P
[43]李薪宇,吕炳朝.暂态混沌神经网络中的模拟退火策略优化.计算机应用.2005,25(10):2410-2412页
[44]谭营,王保云,何振亚.一种具有暂态混沌和时变增益的神经网络及其在优化计算中的应用.电子学报.1998,25(7):818-822页
[45]张强,马润年,许进.一种混沌神经网络模型及其在优化中的应用.系统工程与电子术.2002,24(2):48-50页
[46]谢传泉,何晨.混沌神经网络模型中的模拟退火策略.上海交通大学学报.2003,37(3):36-39页