【作者】 王毅；

【导师】 杨绿溪；

【作者基本信息】 东南大学 ， 信息与通信工程， 2016， 博士

【摘要】 为了应对无线通信数据量的迅猛增长以及智能接入终端数量的大幅增加,业内己开始着眼于面向2020年之后的第五代移动通信系统(5G)的技术研究以及标准化制定工作,尤其是针对5G系统重新定义了峰值速率、用户体验速率、移动性、端到端时延、连接数密度、流量密度、网络频谱效率和能量效率等八大关键能力指标。为了在有限的时频资源下实现这些指标要求,5G系统引入了多种能进一步挖掘频谱效率潜力的超高效能无线传输技术。其中,基于大规模多输入多输出(Massive MIMO,或称为Large-scale MIMO)的无线传输技术,通过大数量天线的使用,深度挖掘空间维度资源,获得了诸多不同于传统MIMO系统的传输特性和物理特性,从而可在不增加时频资源开销的情况下使频谱效率和功率效率在4G系统的基础上再提升一个量级。因此,大规模MIMO技术也被业内普遍认为是5G系统的核心技术之一。然而,大规模MIMO技术在应用过程中仍存在许多有待解决的问题,如频分双工(FDD)制式下适度的导频开销方案及设计、高能效的绿色大规模MIMO资源分配、异构网络大规模MIMO以及分布式大规模MIMO系统的传输容量性能分析和移动场景下时变信道特性对于大规模MIMO系统的性能影响等。基于此,本论文开展了 "面向B4G/5G的大规模MIMO无线通信系统关键技术研究"这一课题,具体研究内容和主要贡献如下:1.针对大规模MIMO FDD下行系统,在普适空间相关信道下分析了导频序列长度随天线数以不同速率变化时的系统遍历可达速率渐进性能,得出可达速率关于导频长度的解析表达式,并由此推导出了满足可达速率持续增长时的导频长度与天线数的数学关系式。推导结果表明:(1)即使在归一化导频长度(即导频长度与天线数之比)趋于0时,只要天线数持续增加,即可保证系统速率无限增加,但其增加速度受到归一化导频长度的影响;(2)若固定导频长度,只增加天线数,则系统速率将很快出现平台效应。进一步,在给定导频序列长度的条件下,提出一种基于可达速率最大化的导频设计方案。该方案将导频结构对于信道估计的作用直接反映于系统可达速率,而非单一的考察信道估计的均方误差性能。仿真结果验证了所作推导的正确性以及所提出的最优导频方案的性能增益。2.研究了大规模MIMO FDD系统的下行信道估计与数据传输两个阶段的联合资源分配问题,在保证信道估计阶段导频时长和功率开销的同时,兼顾数据发送的时长和功率分配。以最大化能效为设计目标,并考虑最小频谱效率要求和总能耗约束,建立关于导频时长、导频功率和数据功率的联合优化模型,提出一种次优的低复杂度三层迭代能效优化算法。由于目标函数形式复杂,首先利用确定性等价近似推导出能效目标函数关于导频长度、导频功率以及数据功率的解析表达式;在此基础上,利用分数规划将原目标函数转换为带参数的等价减式形式,再利用频谱效率下界对转换后目标函数进行缩放,将原非凸问题逐步释放为凸优化问题,最终提出一种三层迭代优化算法对其进行求解。同时,给出了该算法的复杂度分析和收敛性证明。最后,针对两种特殊信道相关阵模型,利用Lambert W函数分别求得了最优数据功率和最优导频时长的闭合形式解。通过数值仿真,验证了不同的信道相关性强弱条件下所提出的迭代算法取得的能效性能增益,并且可以看到本算法具有较快的收敛速率。3.针对多对用户大规模MIMO中继系统,研究了理想信道状态信息(CSI)和含估计误差信道状态信息情况下,不同中继预编码方案的系统频谱效率性能和功率缩放律。首先,推导了系统平均频谱效率关于任意中继天线数、用户数、发射功率向量以及信道统计量的解析表达式。从推导结果可以看到:(1)理想CSI下,最大比发送(MRT)和迫零(ZF)方案对应的频谱效率均随着中继天线数呈近似对数比例增长关系,与干扰用户对个数呈近似对数反比例关系。(2)非理想CSI下,频谱效率与天线数依然满足上述关系,但由于信道估计误差引入了更多的用户间干扰,因而其与用户对个数的变化关系更为复杂。当用户对个数固定时,给出了频谱效率的渐进性能表达式。可以看到在理想CSI和非理想CSI下,信源用户和中继发射功率可同时以最大1/N和最大1/√N倍缩放(N为中继天线数),且保证频谱效率维持在固定水平。此外,还分析了当用户对个数随天线数等比例增长时,系统的频谱效率渐进性能。结果表明,信源节点可获得最大1/√N倍的发射功率缩放增益,但此时中继无法获得功率缩放。这也意味着随着中继天线数增长,系统可以服务更多的用户,同时降低信源发射功率,且通过调整用户对个数和天线数的比值,即可满足用户的任意频谱效率要求,这对于5G密集用户分布场景是极为有利的。仿真结果验证了所推导的频谱效率闭合表达式的准确性以及所获得的发射功率缩放增益范围,并对比了两种预编码方案下的性能变化趋势。4.研究了高能效大规模MIMO中继系统的资源联合分配问题。大规模天线阵列虽然可以大幅降低发射功耗,但却使射频电路功耗成倍增加,从而系统的能效性能受到较大影响,由此对大规模MIMO中继系统进行了基于能效最大化的发射功率和天线数联合优化。对于MRT预编码方案,根据能效目标函数的性质,分别证明了其关于功率向量和天线数的拟凹特性,并由此得到了:(1)天线数一定时的最优信源功率和最优中继功率的数学关系式;(2)在功率向量一定时的最优天线数闭式解。由此,提出一种一维遍历搜索交替迭代算法。由于该方法的收敛速度和性能受初始值和步长的影响较大,为了加快收敛速度,利用上述最优发射功率关系并结合分式规划性质,给出了一种具有超线性收敛速率的基于大信噪比区间近似的交替迭代优化算法。在ZF预编码方案下,首先利用分数规划将目标函数等价转换为带参数的减式形式,进而利用目标函数的部分凸性,从而直接求解得到了最优发射功率和最优天线数的闭合形式解。更为重要的是,所提出的三个算法都只用到了信道统计量和标量计算,且无需信道瞬时信息和大维矩阵运算,大大降低了实际系统中的计算复杂度。仿真结果验证了所提能效功率分配算法的有效性,并展示了所提算法在较少的迭代次数下便可收敛到最优解。5.针对分布式大规模MIMO系统,研究了移动场景中普适时间相关信道下的频谱效率性能和功率缩放律。首先,利用高斯马尔科夫过程对时变信道特性进行建模,以时间相关性系数表征时变快慢程度。其次,利用大维随机矩阵理论以及Gamma分布随机变量特性,推导了单小区分布式大规模MIMO系统在采用MRC接收和MRT预编码时的包含有信道时变系数的上下行频谱效率闭合表达式。进一步,在多小区分布式大规模MIMO系统受到导频污染和信道时变的双重影响下,推导了采用MRC接收机时的上行频谱效率闭合表达式。最后,分别给出了单小区和多小区系统中在总天线数与用户数之间的比值趋于无穷大时的频谱效率渐进解析表达式。从渐进结果中可以看到:(1)在单小区场景中,虽然频谱效率极限值随时间相关系数单调下降,但与无信道时变性时相比,并不影响此时所获得的发射功率缩放增益:(2)多小区中频谱效率极限值与信道时变系数无关,也即频谱效率极限值不受信道时变性影响,且此时所获得的功率缩放增益依然与无信道时变特性时一致。仿真结果表明,所推导的闭合表达式即使在天线数较少的情况下也能很好地逼近真实值,而且在总的发送天线数相同时,时变信道下分布式大规模MIMO系统相较集中式大规模MIMO系统具有更好的性能优势。更多还原

【Abstract】 To meet the challenge of significant and continuing growth of wireless communication traffic and intel-ligent terminals,the industry has embarked researching and standardizing of the 5th generation(5G)mobile communication system that is to be commercialized in the year beyond 2020.Especially,the industry has redefined eight key indexes which include user experience rate,mobility,end-to-end latency,connection density,traffic density,network spectral and energy efficiency and so on.In order to satisfy these require-ments by virtue of limited time and frequency resource,superhigh effective wireless transmission techniques are introduced to 5G which can further excavate the potential of improving spectral efficiency.Among these,the massive MIMO based wireless transmission technology has greatest chance to improve the spectral and power efficiency by order of magnitude without any extra cost of time and frequency overhead.By utilizing a large number of antennas,massive MIMO exploits the spatial resource deeply and brings many transmission and physical characteristics,which are different from conventional MIMO system.Thus,massive MIMO has been widely considered as one of the most sought-after and disruptive technologies for the 5G wireless com-munication systems.However,to facilitate the massive MIMO technology,there are still many open issues that need to be solved,such as,appropriate pilot overhead scheme and pilot design for FDD massive MIMO systems,high energy-efficiency based resource allocation for green massive MIMO,capacity analysis of the heterogeneous massive MIMO and distributed massive MIMO systems,the effects of channel time-varying characteristic on the massive MIMO systems with user mobility and so on.Based on these,we conduct the topic "Research on Key Technologies of Massive MIMO Wireless Communication System for B4G/5G",the main contributions are listed as follows:1.For a massive MIMO downlink frequency division duplex(FDD)system over general correlated Rayleigh fading channel,we investigate the impacts of the pilot sequence length on the ergodic achievable rate,which varies with the number of base station antennas at different speed,and then derive the analytical expression of achievable rate with respect to(w.r.t.)the training sequence length.It is discovered from the analytical results in two-fold that(1)the length of training sequence normalized by the antenna number approaches to zero yet the system capacity is guaranteed to be positive infinity as long as the antenna number is large enough;(2)the transmission capability saturates to a certain level if the antenna number grows to very large with any given training length.Furthermore,for a given training sequence length,we propose an optimal pilot design based on the achievable rate maximization,which combines the effect of channel estimation into the final performance of rate.Simulation results verify the theoretical derivations and the performance gain of our proposed optimal pilot structure.2.The energy efficient resource allocation problem is investigated for the downlink massive MIMO FDD system,which considers the channel estimation and data transmission simultaneously.Our objective is to maximize energy efficiency(EE)by adjusting the training duration,training power and data power,under the constraint of total transmit energy and spectral efficiency requirement of the user.Due to the complicated form of the established problem,the deterministic equivalent approximation methodology is introduced to obtain an analytical expression of the cost function.Based on this,the optimization problem in non-convex fractional form is transformed into an equivalent optimization problem with parametric subtractive form by using fractional programming.Then,with the lower bound of spectral efficiency,the transformed objective function is relaxed to a concave form and finally a 3-layer iterative resource allocation algorithm is proposed to solve it.Moreover,we obtain the closed-form solutions using the Lambert W function for some special channel cases.Numerical results validate the benefits of the proposed scheme and illustrate the tradeoff between training and data transmission.3.We investigate the spectral efficiency(SE)and power scaling laws for a massive MIMO relaying sys-tem with multi-pair users under different CSI conditions,when the relay node adopts different processing schemes.Firstly,we derive the closed-form expression of the SE,based on which,it is shown that(1)For perfect CSI,the SE per user increases with the number of relay antennas but decreases with the number of user pairs,both logarithmically,with MRT and ZF precoding.(2)For imperfect CSI,the SE performance w.r.t.the antenna number still holds as above,but it becomes more complicated w.r.t.the user-pair num-ber due to the inter-user interference caused by channel estimation error.For the fixed user-pair number,we deduce the asymptotic performance of SE,which indicates that for ideal and imperfect CSI cases,the transmit power at the senders and the relay can be simultaneously scaled down maximally by 1/N and(?)(N is the number of relay antennas)while the SE can maintain a constant value.Moreover,when the user-pair number increases with the antenna number.it is shown from the asymptotic performance that the user transmit power can be cut down maximally by(?)but the power saving is unavailable at relay.It also means that the number of served user pairs can grow proportionally over the number of relay antennas with arbitrary SE requirement and no extra power cost.All the analytical results are verified through the numerical simulations and the performance comparison between MRT and ZF precoding schemes is also conducted.4.We consider the EE-oriented resource allocation problem in multi-pair massive MIMO relaying systems.Although the transmit power at the users and relay can be scaled down as the number of relay antennas increases,the circuit power consumption of the radio frequency(RF)chains grows by times,which leads to serious influence on the system energy efficiency(EE).Thus,the relay antenna number and the transmit power at the users and the relay are jointly optimized to maximize the EE.When the MRT precoding is exploited at the relay,the quasi-concavity of the cost function w.r.t.each variable is demonstrated by ex-ploring the properties of objective function,and then we obtain two important results,namely,(1)With the antenna number fixed the mathematical relationship between the optimal transmit power at the users and the relay is given;(2)With fixed transmit power vector the closed-form solution of the optimal antenna number is derived.Accordingly,an one-dimensional exhaustive iterative algorithm is proposed,while the convergence rate and the performance are sensitive to initialization point and the value of step length.To improve the convergence speed,we further adopt fractional programming combining with the first results above to propose a more efficient iterative algorithm with superlinear convergence rate.When the ZF pre-coding is employed at the relay,we first transform the original EE function into an equivalent optimization problem in a parametric subtractive by using fractional programming.Then,we explore the partial con-vexity of the cost function w.r.t.the variables and finally,the closed-form solutions are derived by means of the standard convex theory,i.e.,KKT conditions.What’s more,the proposed algorithms avoid the com-plicated matrix calculations and the requirement for instantaneous CSI of small-scale fading.Therefore,they are computationally efficient in the practical system.Numerical results verify the effectiveness and superiority of the proposed algorithms,which have fast convergence and achieve near-optimal EE.5.We study the SE performance and power scaling laws for a multiuser distributed massive MIMO system under the mobile scenario with the impact of general time-varying channel.Firstly,the first order Gauss-Markov process is introduced to model the time-varying channel,where temporal correlation coefficient was adopted to characterize time-varying degree.Secondly,using the deterministic equivalent principles and the properties of Gamma random variables,we derive the closed-form expression of the uplink and downlink SE including the temporal correlation coefficient in it,when the MRC detection and the MRT precoding are employed.Furthermore,considering the influences of both pilot contamination and channel time variance,we derive the closed-form expression of SE for a multi-cell distributed massive MIMO system with MRC receiver.Finally,the limiting performances of SE are given when the ratio of the total number of antennas to the number of users goes to infinity,which indicate that(1)In the single-cell scenario,the limiting SE decreases with temporal correlation coefficient without affecting the transmit power gain in comparison to the time-invariant channel.(2)In the multi-cell scenario,the limiting SE has no relation to the temporal correlation coefficient and the power scaling law is the same as that of time-invariant channel.Numerical simulations validate the accuracy of the proposed analytical expression of the SE even for small number of antennas,and show that distributed massive MIMO outperforms the centralized massive MIMO under the channel temporal correlation.更多还原