量子遗传算法优化神经网络及其在MIMO系统信号检测中的应用研究
Research on Neural Network Optimized by Quantum Genetic Algorithm and Its Application to Signal Detection of MIMO Systems 主 题 词：量子计算；量子遗传算法；神经网络；多输入多输出；正交频分复用；信号检测
Keywords: Quantum Computation; Quantum Genetic Algorithm; Neural Network; MIMO; OFDM; Signal Detection 摘 要
量子信息学是一门新兴的交叉学科，它在信息领域中有着独特的功能，在提高运算速度、确保信息安全、增大信息容量和提高检测精度等方面可突破现有经典信息系统的极限。特别是近年来，基于量子并行计算的量子智能算法有效地降低了一些经典难解算法的计算复杂度。
在目前的通信系统中，低误码率和低计算复杂度是所有检测技术追求的目标。本文正是基于这一目标，设计新型的优化检测算法，以期达到检测性能和计算复杂度的良好折衷。
本文首先研究了量子遗传算法的特性，并利用量子遗传算法种群规模小，收敛速度快的特性来优化传统神经网络，设计了量子遗传算法优化的BP(Back Propagation)网络和RBF(Radial Basis Function)网络，并分别对这两种量子遗传算法优化的神经网络进行性能测试，实验表明，用量子遗传算法优化过的神经网络在网络性能上比传统神经网络有较明显的提高。
其次，将量子遗传算法优化神经网络应用于MIMO系统信号检测。用量子遗传算法优化神经网络检测信号的初始值，提出了基于量子遗传算法优化神经网络的MIMO系统信号检测方案。实验表明，基于量子遗传算法优化神经网络的MIMO检测方案在检测性能上比基于传统神经网络和基于量子遗传算法的MIMO检测方案有较明显的提高。
最后，研究了基于量子遗传算法优化神经网络的MIMO-OFDM系统信号检测，用量子遗传算法优化神经网络检测信号的初始值，提出了一种基于量子遗传算法优化神经网络的MIMO-OFDM系统信号检测方案。实验表明，基于量子遗传算法优化神经网络的MIMO-OFDM检测方案，其检测性能优于基于传统神经网络和基于量子遗传算法的检测方案。
关键词：量子计算；量子遗传算法；神经网络；多输入多输出；正交频分复用；信号检测
ABSTRACT
Quantum information science is a rising cross subject. Due to unique features in the information field, it may break the limitation of classic information system, be currently available in several aspects, namely, speeding computation, ensuring information security, expanding the capacity of information, improving the accuracy of detection. Particularly in recent years, quantum algorithms of intelligence, based on the parallel quantum computation, effectively simplify computation complexity belonging to some classic algorithms which are not easy to solve problem on the background of classic system.
Low error rate and reduced complexity of communications system are two ultimate targets for all detection techniques. In order to achieve the goals, the dissertation designs a new type of optimization detection algorithm which is expected to acquire desired performance-complexity trade-off.
First of all, the dissertation investigates the characteristic of Quantum Genetic Algorithm (QGA). It makes use of QGA, which has features of small population size and fast convergence, to optimize BP(Back Propagation) network and RBF(Radial Basis Function) network. Simulation results show that the neural networks optimized by QGA perform well in the test.
Secondly, the dissertation discusses the signal detection scheme with neural network optimized by QGA in the MIMO systems. It takes advantage of QGA to optimize the initial data of the neural network. Simulation results show the superiority of the proposed method in MIMO signals detection.
Finally, the dissertation investigates the signal detection scheme with neural network optimized by QGA in the MIMO-OFDM systems. It takes advantage of QGA to optimize the initial data of the neural network. Simulation results show that the proposed method is superior to the other algorithms in MIMO-OFDM signals detection.
Keywords: Quantum Computation; Quantum Genetic Algorithm; Neural Network; MIMO; OFDM; Signal Detection1008
[1] [2] [3] [4] [5] [6] [7] [8] [9] [10] ... 下一页 >>