【作者】 钱雨；

【导师】 王文博；

【作者基本信息】 北京邮电大学 ， 信号与信息处理， 2007， 博士

【摘要】 移动通信的高速发展，以及移动网络和互联网的加速融合，使得移动网络宽带化的需求日益迫切。在这种需求的驱动下，第三代合作伙伴计划(3rd Generation Partnership Proiect，3GPP)在第三代移动通信(3rd Generation，3G)的基础上，提出了其长期演进系统(LongTerm Evolution，LTE)，以提供更高的带宽、更低的时延，以及更好的服务质量(Quality of Service，QoS)保障。3G LTE最重要的改进在于采用全新空中接口技术，其中单载波频分多址(Single Carrier-Frequency Divistion Multiple Access，SC-FDMA)和虚拟多入多出(Virtual Multiple Input Multiple Output，Virtual MIMO)是LTE上行链路中最重要的两种新型技术。SC-FDMA技术不仅实现简单，而且能够有效克服无线通信信道的多径效应引起的符号间干扰。相比正交频分多址(Orthogonal Frequency DivisionMultiple Access，OFDMA)技术，采用SC-FDMA技术的终端具有较低的发射峰均功率比(Peak-to-Average Power Ratio，PAPR)。同时，虚拟MIMO技术则可以在充分利用现有频谱资源基础上，大大增加系统信道容量。当采用这两种技术时，系统具有空、时、频三维可用资源，所以可以更充分地利用无线信道的时变衰落特性和多用户分集增益，通过先进的无线资源管理技术，提高系统容量，满足各个用户的QoS需求。因此，本文把针对这两种技术的无线资源管理作为研究的重点，在如下四个方向提出了相应的技术方案：方向1：对于SC-FDMA系统，邻小区共道干扰是限制其性能的瓶颈之一，本文提出了一种抑制邻小区干扰的分布式功控调度算法。各小区基站独立地进行用户调度和功控，从而避免了多小区联合调度带来的高复杂度。仿真结果表明该算法可以降低小区间干扰，提高系统吞吐量，并且可以提供一定的公平性保障。方向2、3：当采用虚拟MIMO时，如何利用用户配对所得虚拟MIMO信道矩阵特性，增大系统吞吐量，是提升系统性能的关键。在虚拟MIMO无线资源管理中，用户配对调度有2种方式：1)根据与虚拟MIMO信道无关的某种调度算法先调度一个用户，然后选择与该用户配对信道容量最大的用户形成虚拟MIMO，这种方案较为简便，但不便于提供QoS保障。对此，本文提出的算法将待配对用户按照信噪比(Signal To Noise Ratio，SNR)高低分为两组，各组有自己相应的配对准则，用户配对优先在SNR相近的用户组内进行，同时，提供跨组配对边际效用函数，以便最大化利用多用户分集。仿真结果表明，该算法在提供较高系统吞吐量的同时，为低SNR用户提供了一定的公平性保障。2)调度算法与虚拟MIMO配对结合，在所有可能的用户配对方案中，寻找配对后效用函数最大的用户组合进行调度，这种方案便于和QoS保障机制结合，但是实现相对复杂。对此，本文结合3G LTE中实时业务QoS时延需求，提出了一种将用户无线链路控制(RadioLink Control，RLC)层待传队列时延和物理层SC-FDMA、虚拟MIMO信道衰落特性联合考虑的跨层调度算法。仿真结果表明，该算法既能保障用户QoS时延需求，又能提供一定的公平性。方向4：当采用方向3类型的虚拟MIMO调度时，遍历查找最优用户配对组合计算复杂度非常高。本文提出了一种降低调度复杂度的改进方案。首先求解各用户在单输入多输出方式下的调度因子，并将调度因子较大的用户组成一个群体。然后以此群体作为优选初始群体，利用遗传算法寻找最佳配对组合；或者直接在此群体内遍历搜索最优用户配对组合。仿真结果表明，利用遗传算法方案，可以取得逼近全局遍历搜索的调度效果；而局部群体遍历搜索算法虽然调度精度低于遗传算法，却可以进一步降低计算复杂度。实际系统中可以根据精度和计算复杂度需求，选择相应的方案。更多还原

【Abstract】 Rapid development of mobile communication and the merging of mobile network and Internet have put dramatically growing demands for the wide-band mobile communication. Therefore, the 3^rd Generation Partnership Project （3GPP） has proposed a new mobile communication system evoluted from the 3^rd Generation system, which is called as the 3G long term evolution （3G LTE）. It will provide higher bandwidth, lower latency, and better quality of service （QoS） guarantees.The most significant improvement of 3G LTE uplink system is that it employs many novel air-interface techniques. Among them, the single carrier-frequency division multiple access （SC-FDMA） and virtual multiple input multiple output （Virtual MIMO） are one of the most important ones. SC-FDMA is not only easy to be realized but also can overcome the inter-symbol interference due to the multi-path mobile channel propagation. Compared to orthogonal frequency division multiple access （OFDMA）, SC-FDMA has the lower peak-to-average power ratio （PAPR） at the transmitter. On the other hand, the Virtual MIMO can increase more channel capacity than the single input multiple output （SIMO） significantly.When employing these two techniques, the schedulable resource of the system becomes three dimensions including space, time and frequency domain. So more multi-user diversity gain can be exploited by utilizing the time-variant channel fading. Thus the system capacity can be further improved by the advanced radio resource management （RRM） technologies and the QoS of each user can also be guaranteed. Therefore, this dissertation focuses on the radio resource management based on these two techniques. There are four major schemes proposed in this dissertation.Scheme 1: For SC-FDMA system, the inter-cell co-channel interference is the bottleneck of the system capacity. A distributed power control scheduling algorithm is proposed to solve this problem. In this algorithm, the scheduling and power control are carried out in each cell independently in order to avoid the complexity of joint multi-cell scheduling. The simulation results show that this algorithm can reduce inter-cell interference and improve the system capacity with providing the reasonable fairness.Scheme 2 and 3: When employing Virtual MIMO, how to pair users to acheive high Virtual MIMO channel capacity is the key to improve the system performance. In Virtual MIMO, there are two types of user paring scheduling for the radio resource managements:Type 1: The system schedules one user, according to a traditional scheduling algorithm which is not related with Virtual MIMO channel, and then chooses another pairing user to maximize the Virtual MIMO channel capacity. This type is simple, but it can’t provide enough QoS guarantees. A pairing algorithm is proposed in this dissertation based on signal to noise ratio （SNR） of the users. The users are divided into two groups, i.e. high-SNR group and low-SNR group, each group has its individual pairing metric. The co-group users of the scheduled user have the pairing priority, and a marginal utility function is provided for crossover-group pairing. The simulation results show that the algorithm can provide higher system throughput and maintain fairness for low SNR users.Type 2: The scheduling algorithm is combined with Virtual MIMO user pairing; the system chooses the user pair, whose utility function value is the biggest, from all possible user pair candidates. This type can provide QoS guarantees, but it is complex. For this case, a cross-layer scheduling algorithm is proposed. The algorithm utilizes the queue information from radio link control （RLC） layer, and the channel fading information of physical layer. The simulation results show that the algorithm can guarantee the QoS latency request of users, and provides fairness in a manner.Scheme 4: In Virtual MIMO scheduling type 2, the computing complexity of exhaustive searching best user pair is very high. In this dissertation, the scheduling factor of users in SIMO mode are calculated, and the scheduler chooses some users whose scheduling factor are big to form a group. Then, the scheduler uses this group as the initial population of the genetic algorithm to find the best pair. Another method is to search best pair in this local group directly. The simulation results show that the performance ot the genetic scheme with the chosen initial population, can approach the performance of the exhaustive search scheme. As for the local group searching scheme, though the scheduling precision is lower, it can further reduce the computing complexity. So they can be chosen to use according to the request of complexity and precision of scheduling.更多还原