非完善CSIT下MIMO系统能效优化方法研究

发布时间：2018-09-19 12:53

【摘要】：能量效率(Energy Efficiency,EE)已经成为第五代移动通信系统(5G)的关键性能指标(Key Performance Indicator,KPI)之一。提高无线通信系统的能量效率能够有效降低移动网络运营商(Mobile Network Operator,MNO)的基础建设费用(Capital Expenditure, CAPEX)和运营费用(Operation Expenditure, OPEX),同时也能够大幅度降低信息通信技术(Information Communication Technology,ICT)的温室气体排放量,实现低碳社会的目标。另一方面,多输入多输出(Multiple Input Multiple Output,MIMO)技术已经成为包括5G在内的很多无线通信标准的关键技术,极大的提高了频谱效率(Spectral Efficiency, SE)。考虑到实际的MIMO系统的发送端信道状态信息(Channel State Information at Transmitter,CSIT)往往并不完善,因此很有必要针对非完善CSIT下的MIMO系统进行能量效率优化方法的研究。本文关注非完善CSIT下的下行MIMO系统能量效率优化方法的研究。由于预编码技术是MIMO系统获得分集和复用增益的基础,因此本文首先在受限的CSIT不确定度模型下针对多用户MIMO(Multiuser MIMO,MU-MIMO)系统研究了能量效率优化的预编码设计问题。另外,考虑到训练序列开销与能量效率之间存在折中关系,为此本文分别针对基于训练的多用户MIMO系统和多小区MIMO系统研究了能量效率优化方法。首先,预编码技术是提高多用户MIMO系统性能的重要手段。因此,本文在受限的CSIT不确定度模型下研究了能量效率优化的预编码设计问题。该问题是一个非凸的分数规划。为了求解该问题,本文首先运用用户可达速率与最小均方误差(Minimum Mean Square Error,MMSE)的对应关系以及min-max不等式把原问题转化为它的具有分数和max-min形式的下界问题。接着利用分数规划定理把分数形式的问题转化为带参数的、减数形式的优化问题,并利用拉格朗日对偶性将该max-min问题转化为max-max的形式,进一步提出了能够保证收敛性的迭代求解算法。仿真结果表明,本文所提出的算法相比于已有的算法能够显著提高多用户MIMO系统的能量效率。其次,基站获得CSIT需要付出一定的训练序列开销,训练序列功率过小会导致CSIT准确度很低,造成能量效率的下降；训练序列功率过大则会导致速率增益不能补偿训练序列的能耗,同样也会降低能量效率,因此,在训练序列功率和能量效率之间存在折中关系,需要仔细研究。本文针对基于训练的多用户MIMO系统研究了能量效率优化方法,形成了能效优化的功率分配问题。针对CSIT不可见的事实提出一种两步优化方法,分别优化各态历经能量效率和瞬时能量效率。在各态历经能量效率的优化中,本文首先推导获得了各态历经可达速率的更紧致下界,然后把原问题转化为关于各态历经能量效率下界的优化问题,并证明了它关于训练功率和数据功率是联合拟凹的,进一步提出了一种交替优化求解算法。在瞬时能量效率的优化中,发送端采用基于各态历经能量效率优化得到的训练序列功率发送训练序列,接收端进行信道估计得到有误差的信道状态信息,并据此预测瞬时能量效率,进一步提出了一种用于优化数据信号发送功率的瞬时能量效率优化算法。仿真结果表明,本文所提出的两步能量效率优化算法相比于只优化各态历经能量效率的算法和频效优化的算法均能提高多用户MIMO系统的能量效率。另外,单小区传输(Single Cell Processing, SCP)和多小区协作波束成型(Coordinated Beamforming, CBF)是多小区MIMO系统中两种典型的传输模式,考虑到用户可能处于不同的位置,采用不同传输模式会带来不同的能量效率。因此,本文形成了基于训练的多小区MIMO系统中考虑最小速率限制的能量效率优化问题。该优化问题可以表示成关于数据信号功率、训练信号功率、传输模式的多变量混合分数规划。本文提出了各态历经能量效率的近似表达式,然后基于该近似表达式提出了优化发送功率的两步算法。第一步,忽略最小速率限制,提出一种交替优化算法来求解无约束的能量效率优化问题；第二步,判断第一步的解是否满足最小速率限制条件,如果满足则算法结束,否则将考虑最小速率约束的问题转化为最小化功耗的问题并提出一种线性规划算法进行求解。这两个算法都能够保证收敛性。在优化发送功率之后,本文提出采用遍历搜索的方法得到能效优化的传输模式。仿真结果表明,所提出的能量效率优化算法相比于一直以最大功率进行发送的频效优化的基准算法在最小速率限制较低时能够获得能量效率的较大增益。而最小速率限制条件也会影响到最优传输模式的选择,限制条件越苛刻,系统越倾向于采用协作波束成型传输模式。原因在于：基站需要提高发送功率以满足较高的最小速率限制条件,增大了小区间干扰(Inter-Cell Interference, ICI),使得系统能量效率受限于小区间干扰,因此,基站就越倾向于采用协作波束成型传输模式消除小区间干扰,提高多小区系统的能量效率本文首先在受限的CSIT误差模型下针对多用户MIMO系统提出了能效优化的预编码设计方法,然后分别在基于训练的多用户MIMO系统和多小区MIMO系统中提出了能效优化方法。本文研究结果对于如何提高非完善CSIT下的MIMO系统能量效率具有重要的参考价值。
[Abstract]:Energy Efficiency (EE) has become one of the key performance indicators (KPIs) of the fifth generation mobile communication systems (5G). Improving the energy efficiency of wireless communication systems can effectively reduce the capital Expenditure (CAPEX) and transportation costs of mobile network operators (MNO). Operational Expenditure (OPEX) can also significantly reduce greenhouse gas emissions from information communication technology (ICT) and achieve the goal of a low carbon society. On the other hand, MIMO technology has become a lot of wireless including 5G. The key technology of communication standard greatly improves the spectral efficiency (SE). Considering that the actual transmitter channel state information (CSIT) of MIMO system is not perfect, it is necessary to study the optimization method of energy efficiency for MIMO system under imperfect CSIT. Study.
This paper focuses on the study of energy efficiency optimization methods for downlink MIMO systems under imperfect CSIT. Since precoding technology is the basis for obtaining diversity and multiplexing gains for MIMO systems, the precoding presupposition of energy efficiency optimization for multiuser MIMO (MU-MIMO) systems under constrained CSIT uncertainty model is studied in this paper. In addition, considering the tradeoff between training sequence overhead and energy efficiency, this paper studies the energy efficiency optimization methods for training-based multi-user MIMO systems and multi-cell MIMO systems.
First of all, precoding technology is an important means to improve the performance of multi-user MIMO systems. Therefore, this paper studies the precoding design problem of energy efficiency optimization under the constrained CSIT uncertainty model. This problem is a non-convex fractional programming. To solve this problem, user reachable rate and minimum mean square error (Minim) are first used. The corresponding relation of UM Mean Square Error (MMSE) and min-max inequality transform the original problem into its lower bound problem with fractional and max-min forms. Then the fractional form problem is transformed into a parametric, minus form optimization problem by using the fractional programming theorem, and the max-min problem is transformed into a max-min problem by using Lagrange duality. In the form of - max, an iterative algorithm is proposed to guarantee the convergence. Simulation results show that the proposed algorithm can significantly improve the energy efficiency of multi-user MIMO systems compared with the existing algorithms.
Secondly, the base station needs to pay a certain amount of training sequence overhead to obtain CSIT, the training sequence power is too small will lead to low accuracy of CSIT, resulting in a decrease in energy efficiency; training sequence power is too large will lead to the rate gain can not compensate for training sequence energy consumption, also will reduce energy efficiency, therefore, in the training sequence power and energy efficiency. There is a trade-off between rates, which needs careful study. In this paper, the energy efficiency optimization method is studied for training-based multi-user MIMO system, and the power allocation problem of energy efficiency optimization is formed. In the optimization of energy efficiency, a more compact lower bound of the ergodic reachable rate of each state is derived firstly, and then the original problem is transformed into an optimization problem about the lower bound of ergodic energy efficiency of each state. It is proved that the training power and data power are joint quasi-concave, and an alternating optimization algorithm is proposed. In the optimization of instantaneous energy efficiency, the transmitter uses the training sequence power transmission training sequence based on ergodic energy efficiency optimization, and the receiver gets the channel state information with errors by channel estimation, and predicts the instantaneous energy efficiency accordingly. A new instantaneous energy used to optimize the transmission power of the data signal is proposed. Simulation results show that the two-step energy efficiency optimization algorithm proposed in this paper can improve the energy efficiency of multi-user MIMO systems compared with the algorithms that only optimize ergodic energy efficiency and frequency efficiency optimization.
In addition, SCP (Single Cell Processing) and CBF (Coordinated Beam Forming) are two typical transmission modes in multi-cell MIMO systems. Considering that users may be in different locations, different transmission modes will bring different energy efficiency. This optimization problem can be expressed as a multivariate mixed fractional programming with respect to data signal power, training signal power and transmission mode. An approximate expression of ergodic energy efficiency is presented in this paper. Based on this approximate expression, the optimal transmission is proposed. In the first step, an alternative optimization algorithm is proposed to solve the unconstrained energy efficiency optimization problem, ignoring the minimum rate constraint. In the second step, it is determined whether the solution of the first step satisfies the minimum rate constraint condition, and if it satisfies the minimum rate constraint, the algorithm is terminated. A linear programming algorithm is proposed to solve the problem of energy consumption. Both algorithms can guarantee the convergence. After optimizing the transmission power, an ergodic search method is proposed to optimize the transmission mode of energy efficiency. The benchmark algorithm for frequency-efficiency optimization can achieve greater energy efficiency gains when the minimum rate constraint is low. The minimum rate constraint also affects the selection of the optimal transmission mode. The more severe the constraint conditions are, the more likely the system is to adopt cooperative beamforming transmission mode. Higher minimum rate constraint enhances inter-cell interference (ICI), which limits the energy efficiency of the system to inter-cell interference. Therefore, the base station is more inclined to adopt cooperative beamforming transmission mode to eliminate inter-cell interference and improve the energy efficiency of multi-cell system.
In this paper, a precoding design method of energy efficiency optimization for multi-user MIMO systems based on constrained CSIT error model is proposed firstly, and then energy efficiency optimization methods are proposed for training-based multi-user MIMO systems and multi-cell MIMO systems respectively. Important reference value.
【学位授予单位】：中国科学技术大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：TN919.3

【参考文献】