修正的BFGS法研究及其在MIMO信号检测中的应用.pdfVIP

摘 要 本文在研究经典拟牛顿BFGS(Broyden, Fletcher, Goldfard, Shanno)算法的基础 上，提出了一种关于矩阵变量的修正BFGS算法，研究了该算法在MIMO(Multiple- Input-Multiple-Output)信号检测中的应用。主要内容如下： 1.介绍了非精确步长搜索Wolfe-Powell准则，给出了经典BFGS算法原理、全局 收敛性分析以及超线性收敛性分析。 2.提出了修正BFGS算法，给出了修正BFGS算法的数学模型，全局收敛性分析 以及超线性收敛性分析。针对二次凸问题，将修正BFGS算法与经典BFGS算法进 行了仿真对比分析。 3.分别采用经典BFGS算法与修正BFGS算法，结合不动点方程求解MIMO信号 检测问题。仿真结果表明，修正BFGS算法能够降低计算复杂度，提高计算效率， 减少计算时间。 关键词： BFGS 全局收敛性 超线性收敛性 MIMO 信号检测 Abstract Based on the analysis of the classical BFGS algorithm, this thesis proposes a modified BFGS algorithm and focuses on its application in MIMO detection problem. The main contents are as follows: 1. Introduce the inexact line search method of Wolfe-Powell and mathematical structure of the classical algorithm with its global convergence analysis and superlinear convergence analysis. 2. Based on the classical BFGS algorithm, we propose the modified BFGS algorithm with its mathematical structure, convergence analysis and superlinear convergence analysis. Compare the classical BFGS algorithm with the modified BFGS algorithm by the two-order convex problem through simulation. 3. Simulate the MIMO detection problem through classical BFGS algorithm and modified BFGS algorithm with the fixed point equation. Simulation results show that, the modified BFGS algorithm can reduce the computational complexity, improve the calculation efficiency and reduce the computation time. Keywords: BFGS Global convergence Superlinear convergence MIMO detection 目 录 第一章 绪论 1 1.1 引 言 1 1.2 课题研究的目的及意义2 1.3 本文内容安排3 第二章 经典BFGS 算法及其收敛性分析5 2.1 非精确步长搜索准则5 2.1.1 Wolfe-Powell 准则5