导　　师： 李艳萍

学科专业： H1001

授予学位： 硕士

作　　者： ;

机构地区： 太原理工大学

摘　　要： 众所周知，码分多址(code division multiple access，cdma)系统采用扩频通信技术，大幅度地提高了频率利用率，具有容量大、覆盖范围广、手机功耗小等突出优点，但是在cdma系统中，由于多个用户使用的扩频码集一般不能严格正交，故存在多址干扰(multiple access interference，mai)，它对系统容量起着制约性的作用。多用户检测技术作为一种性能优越的抗多址干扰技术，已经成为第三代移动通信中的一项关键技术，它在传统检测技术的基础上，联合考虑同时占用某个信道的所有用户或某些用户，能有效地消除或减弱其它用户对期望用户的影响。由于多用户检测问题的实质是组合优化问题，因此，人工神经网络所具有的快速优化计算能力和大规模并行处理能力使其在多用户检测问题中表现出良好的前景。而暂态混沌神经网络(transiently chaotic neural network，tcnn)在hopfield神经网络(hopfield neural network,hnn)基础上引入了一个逐渐消失的自反馈项，使网络在搜索过程中具有比一般神经网络更复杂的暂态混沌动力学特性，在处理组合优化问题上具有更大的优势。因此，本文将暂态混沌神经网络和多用户检测相结合，在原有的基于暂态混沌神经网络多用户检测器的基础上加以改进，提出一种快速收敛的改进自适应暂态混沌神经网络多用户检测器，并将其应用到多载波cdma(multi-carrier cdma，mc-cdma)系统中，达到减小误码率、优化系统性能的目的。 As everyone knows, spreading spectrum communication technology is used in code division multiple access /(CDMA/) system and the frequency utilization has been improved significantly. CDMA system has many advantages such as large capacity, wide covering range, small power and so on. But in CDMA system, the spreading codes of different users are generally notstrictly orthogonal, so there is multiple access interference /(MAI/), whichconstrains capacity of CDMA system. Multi-user detection technology has become a key technology of the third generation mobile communication because of the superior performance of anti-MAI. Based on the traditional detection technology, multi-user detection uses all users' or some users' information occupying a certain channel and can effectively eliminate or weaken other users' impact on the expected user . Since the essence of multi-user detection is combinatorial optimization problem, artificial neural network which has capabilities of rapid optimization and large-scale parallel processing shows good prospect in multi-user detection problem. Transiently chaotic neural network /(TCNN/) introduces a gradually disappearing self-feedback to Hopfield neural network /(HNN/), so that the network has complex dynamics of transient chaos in the search process and has tremendous advantages on processing combinatorial optimization problem. Therefore, this paper combines transiently chaotic neural network and multi-user detection. On the basis of original multi-user detectors based on transiently chaotic neural network, a fast convergence multi-user detector based on the improved adaptive transiently chaotic neural network is presented and it is applied in the multi-carrier CDMA /(MC-CDMA/) system to reduce bit error ratio /(BER/) and optimize system performance. The main work of this paper can be divided into three parts: First, the knowledge of CDMA communication system and multi-user detection technology are introduced; Second, chaos theory and chaotic neural network model are analyzed; Third, multi-user detection technology based on transiently chaotic neural network in mobile communication system is researched and simulated. The third part of this paper is the focus of the work, which mainly includes the following aspects: 1. The optimal multi-user detector based on Hopfield neural network is researched and its energy function is easily trapped by the local minimum points. Since the multi-user detector based on transiently chaotic neural network is analyzed. The simulation results show that to a certain extent, the energy function of multi-user detector based on transiently chaotic neural network can escape the local minimum points. 2. The parameters of transiently chaotic neural network are researched. On basis of existing multi-user detectors based on transiently chaotic neural network, a multi-user detector based on the improved transiently chaotic neural network is proposed and the simulation results show that BER performance of the improved detector is better.than that of existing multi-user detectors based on transiently chaotic neural network. 3. In order to make the network converge faster, the paper improves typical chaos part in transiently chaotic neural network and proposes a fast convergence multi-user detector based on the improved adaptive transiently chaotic neural network. The simulation results show that the fast convergence multi-user detector both has good BER performance and converges faster. 4. The multi-user detector based on transiently chaotic neural network is applied to MC-CDMA system. The simulation results show that the fast convergence multi-user detector based on the improved adaptive transiently chaotic neural network can effectively suppress inter-symbol interference, has the good BER performance and is suitable for data transmission with high rate.

关 键 词： 码分多址系统 扩频通信 频率利用率 混沌神经网络 多用户检测 多址干扰

分 类 号： [TN929.533 TP183]

领　　域： [电子电信] [电子电信] [自动化与计算机技术] [自动化与计算机技术]