【作者】 李倩；

【导师】 王公仆；

【作者基本信息】 北京交通大学 ， 计算机科学与技术， 2018， 硕士

【摘要】 目前,铁路数字移动通信系统(Global System for Mobile communications for Railway,GSM-R)是国内高速铁路最广泛应用的无线通信系统。然而,随着互联网的发展,列车乘客对移动通信的要求已逐渐转变为视频电话、网络多媒体等高宽带服务的形式,GSM-R的窄带特性难以满足列车乘客的数据业务需求。因此,新一代高速铁路无线通信系统应运而生。由于3GPP提出的长期演进(Long Term Evolution,LTE),相对于第三代移动技术(3rd Generation,3G),具有传输速率高、延迟降低、广域覆盖和向下兼容等性能优势。于是,国际铁路联盟拟在GSM-R系统的基础上引入LTE技术加以改进,使其最终演进成LTE-R(LTE-Railway)。一般情况下,普通列车和基站之间的相对运动速度为120km/h,而高铁场景能够高达360km/h,由此导致用户通信伴随着极为严重的多普勒频移现象。LTE下行链路采用正交频分复用(Orthogonal Frequency Division Multiplexing,OFDM)技术,多普勒频偏会破坏子载波间的正交性,引发严重的载波间干扰,进而影响OFDM的系统性能。另外,毫米波是近几年的一大研究热点,能够支持比低频电磁波更高数量级的带宽和容量。若将毫米波应用到高铁场景,必然产生更为严重的多普勒频偏。因此,如何解决多普勒频偏的估计与校正问题对LTE-R的发展和毫米波的推广均具有重大意义。在接收端和发送端相干解调的前提下,多普勒频偏模型与载波频偏模型相似。本文考虑到载波频偏估计技术已经非常成熟,因此将载波频偏估计算法结合LTE运用到了多普勒频偏估计中。文中总结并仿真分析了四种经典的载波频偏估计算法:最大似然估计算法、基于循环前缀的估计算法、Moose算法和SC算法。本文在莱斯信道模型的基础上,根据多普勒频偏的引入方式,对高铁场景分析了两种类型的信道建模。我们假设莱斯信道各分径的到达角相同,从运算时间、估计范围、估计性能三方面对比分析了四种算法在LTE系统的应用性能,探究了高架桥、路堑、隧道三种场景下的性能差异,给出Moose算法在三种场景中估计性能最好且其性能与具体场景无关的结论。最后,我们针对强LOS场景的信道建模,充分挖掘高铁场景下列车在不同时刻经过同一地点的参数之间具有相关性的特征,对Moose算法做了进一步改进。由于强LOS场景下高铁信道只有主径引入多普勒频偏,我们首先通过信道估计提取出接收信号的LOS径信息,然后以历史测量数据作为先验信息提出了一种基于Moose的改进算法。本文在理论推导的基础上,通过仿真实验证明了改进Moose算法比原方法具有更好的估计性能。更多还原

【Abstract】 Nowadays,the global system for mobile communications for railway(GSM-R)is a digital wireless communication system which has been widely used in China.However,with the development of Internet,the requirements of passengers for mobile communication are extended from the original voice communication to high broadband service,including video telephone and online multimedia.GSM-R can not meet the demand of data service from train passengers due to its narrow-band characteristic.Therefore,a new generation of high-speed railway wireless communication system is necessary.The International Union of Railways(UIC)has proposed that 3G technology is not suitable for the railway system,while the Long-Term Evolution(LTE)proposed by 3 GPP has a strong performance advantage over the 3G system in data rate,latency,coverage and downward compatibility.Therefore,LTE is introduced to be the next communication systems for railways,which is often referred to as LTE-R(LTE-Railway).In the high-speed railway scenario,the relative speed between trains and base stations can be as high as 360 km/hour,much larger than the regular train speed 120 km/hour,which gives rise to the severe doppler frequency shift phenomenon.And the orthogonal frequency division multiplexing(OFDM)system is exploited in the downlink of LTE which is sensitive to the doppler frequency shift.The doppler frequency shift can damage the orthogonality of the subcarriers,which will result in serious inter-carrier interference(ICI),and hence affect the performance of OFDM.In addition,the millimeter-wave is a research hotspot in recent years,which can support higher order of bandwidth and capacity.If the millimeter-wave is applied into the high-speed railway,which inevitably will lead to more serious doppler frequency shift.Therefore,how to solve the estimation and correction of doppler frequency shift is of great significance to the development of LTE-R and the promotion of millimeter wave.The doppler frequency offset model is similar to the carrier frequency offset model under the assumption of coherent demodulation operated between the receiver and transmitter.And there have been many methods for carrier frequency offset estimation,which can be applied into the doppler frequency offset eatimation model.Therefore,we firstly conclude four classical carrier frequency offset estimation algorithms:the maximum likelihood estimation algorithm,the estimation algorithm based on cyclic prefix,Moose algorithm and SC algorithm.In the paper,according to the different introduction of doppler frequency shift,two types of channel modeling for high-speed railway are carried out based on the rician channel model.Then we compare and analysis the four estimation algorithms from operation time,estimation range and estimation performance in LTE system.The performance difference is also explored under the viaduct,cutting and tunnel scenario.As a conclusion,the Moose algorithm performs best and has a stable estimation performance in three scenarios.Finally,to further improve the performance of the Moose algorithm for strong LOS scenario,we utilize the correlation between the parameters at different time intervals in the same position on the railway.Due to the doppler frequency offset only has impact on the main path of the channel model in strong LOS scenario,firstly we extracted LOS path information of the signal through the channel estimation.Then the improved algorithm based on Moose is proposed by using the historical measurement information as a prior information.Based on the theoretical derivation,the simulation result shows that the improved Moose algorithm performs better than the original Moose algorithm.更多还原