标题: 基于QR分解之无线通讯多输入多输出侦测
QRD-based MIMO Detection in Wireless Communications
作者: 潘建宏
Pan, Chien-Hung
李大嵩
Lee, Ta-Sung
电信工程研究所
关键字: 多输入多输出侦测;MIMO detection
公开日期: 2011
摘要: 多重输入多重输出技术为一个提供较高频宽效能的有效解决方案。然而,多输入多输出侦测之缺失起因于天线间的干扰效应及ill-condition问题。为了将天线间的干扰效应降低,我们提出基于QR技术的有效率基因演算结合可靠的初始值,三角矩阵型式与多样式突变率。为了对抗ill-condition问题,我们研究于有限回授系统下,基于SV选择法来寻找预编码以最佳化多重输入多重输出侦测效能。为达选择法之最佳化,吾人提出QR选择法是依据有效率分解法与较大对角元素边界值所推导出来的并用于对抗ill-condition问题并将此发展于切换系统与多模预编码系统。新提出方法利用SVD与GMD可预先决定传输模式下来找寻预编码。这些方法主要优势是具备较低计算复杂度与较佳侦测效能。此提出方法不仅适用于单使用者多输入多输出系统,并且适用于多使用者操作多重输入多重输出系统以协调预编码。效能分析与模拟验证所提出QR方法可以使用较低的计算复杂度亦能达到现存方法的效能。
Multiple input multiple output (MIMO) technology is a promising solution to provide higher spectral efficiency. However, the impairment of MIMO detection (MD) is caused by inter-antenna interference (IAI) and ill-conditioned problem. To better mitigate IAI, effective GA-MD associated with QR-based techniques exploit good initial setting, triangular form and diversity-based mutation (DM) are proposed. To prevent the ill-conditioned problem, we study conventional SV-based section criterion to find the precoder to optimize the performance of MD in the limited feedback systems. To optimize the section criterion, the QR-based selection criteria depending on effective decomposition and a larger bound among diagonal entries are proposed to resist the problem developed in switching and multi-mode precoding systems. The new criteria exploiting SVD and GMD perform a predetermined mode to find the precoder. The main benefits of the proposed criteria are their lower computational complexity and better detection performance. The proposed criteria are suitable not only for single user MIMO (SU-MIMO) system, but also for multi-user MIMO (MU-MIMO) system with the cooperative precoding. Performance analysis and computer simulations confirm that the proposed QR-based schemes attain the performance of existing schemes with a significantly lower complexity level.
URI: http://140.113.39.130/cdrfb3/record/nctu/#GT079213819
http://hdl.handle.net/11536/40368
显示于类别:Thesis