基于神经网络预测CQI的LTE下行调度  

A LTE Downlink Scheduling Based on CQI Predicted by Neural Network

作　　者：吴斌[1] 糜正琨[1] 王文鼐[1] 卢元晨 朱泽文[1] 

机构地区：[1]南京邮电大学宽带无线通信与传感网技术教育部重点实验室,南京210003 [2]华为软件技术有限公司,南京210012

出　　处：《北京邮电大学学报》2013年第6期45-49,共5页Journal of Beijing University of Posts and Telecommunications

基　　金：国家自然科学基金项目(60872018);国家科技重大专项项目(2011ZX03005-004-03);江苏高校优势学科建设工程资助项目

摘　　要：提出一种基于神经网络预测信道质量指示(CQI)的长期演进下行调度方法,设计了基于神经网络的CQI预测模型,使用系统中已获得的CQI历史值来预测调度时刻的当前CQI值,调度器使用此预测值进行调度.仿真结果表明,该方法对由CQI延迟所导致的系统性能下降有明显改善.Using the available outdated channel quality indicator （CQI） , an optimized long term evolu- tion downlink resource allocation method is proposed. The key is a neural network based on CQI predic- tion model and algorithm, which can predict the current CQI from the outdated CQIs. The scheduler uses the predicted CQI to allocate radio resources and decide on proper modulation and coding scheme. Simu- lation shows that the proposed prediction model and algorithm have practical effects on solving the degra- dation of scheduling efficiency due to the outdated CQI.

关 键 词：3GPP长期演进 无线资源调度 信道质量指示预测 神经网络 

分 类 号：TN929.53[电子电信—通信与信息系统]