EARS: Future Wireless Broadband Access: Cross-Optimizing Hardware, Physical and Network Layers

提供负担得起且无处不在的宽带连接是繁荣的必要条件，因为宽带无线接入几乎影响到社会和经济的所有部门，包括教育、医疗保健、交通和安全。在美国这样一个幅员辽阔、城乡混杂的国家，通过有线基础设施提供高速宽带数据访问的成本可能很高。另一方面，无线可以非常有效地覆盖大范围和覆盖大量人群。此外，无线是我们连接到互联网并享受娱乐、教育、医疗保健、电子商务、社交网络和远程工作等各种服务的首选媒介。在这种情况下，处理预计将大幅增加的无线数据需求的能力不仅对无线行业至关重要，而且更广泛地对我们的经济增长至关重要。虽然无线连接在过去几十年中有了显著改善，但它远远落后于理论和技术上可实现的限制，而且无法满足未来的需求。考虑到这一点，该项目为具有密集部署的基站的下一代蜂窝网络开发了创新的多层分层基础设施，以及一套集成良好的跨层设计技术，用于干扰管理和系统优化。该方法可以显著提高宽带无线接入的速率性能和用户容量，有望在宽带无线接入的理论和实践之间架起一座桥梁。为了支持无线宽带业务中急剧增长的移动数据流量，本文重点研究了一种基于三大支柱的系统跨层系统优化方法：1)在物理层，使用具有大规模多输入多输出天线系统的基站；2)在无线网络体系结构层面，选择多层异构网络方法，实现前所未有的空间频谱重用；3)在跨层优化层面，提出了一种整体网络效用最大化方法，从全局优化方案的结构出发，系统地获得分层的协议架构。关于上述支柱，本项目将解决的基本挑战是：1)基于混合波束形成的概念，即基于数字基带处理和模拟射频波束形成之间的多用户预编码和小区间干扰管理功能的最优分割，设计集成和功率高效的可重构大规模多输入多输出前端天线系统；2)设计利用长期信道统计来进行小区间协调干扰管理的混合波束形成方案，并利用瞬时信道状态信息在每个小区中实现空间复用增益；3)提出了一种用户划分和调度方法，该方法根据体验质量要求、信道统计和移动性对用户空间进行聚类，为不同的用户群分配网络效用函数，求解组合网络效用最大化问题，并根据优化解的结构特性系统地推导出分层协议体系结构。此外，基于随机几何和随机矩阵理论的先进工具，这项工作将极大地扩展当前无线网络的数学性能分析，以便定量评估所提出的方法相对于现有技术的性能增益。最后，将使用配备了将开发的前端的软件定义的无线电进行小规模实验。

The availability of affordable and ubiquitous broadband connectivity is a necessary condition for prosperity since broadband wireless access impacts virtually all sectors of society and economy including education, healthcare, transportation, and security. In a vast and mixed urban/rural country such as the United States, providing high-speed broadband data access through the wired infrastructure can be costly. On the other hand, wireless can cover large areas and reach a large number of people very effectively. In addition, wireless is the preferred medium through which we connect to the Internet and enjoy a whole wealth of services such as entertainment, education, healthcare, e-commerce, social networking, and remote working. In this context, the ability to handle the predicted dramatic increase of demand for wireless data has become crucial not only for the wireless industry but, more in general, for the growth of our economy. While wireless connectivity has significantly improved over the past few decades, it is quite behind the theoretical and technological achievable limits and it cannot address future demand. With this in mind, this project develops an innovative multi-tier hierarchical infrastructure for next-generation cellular networks with densely deployed base stations, along with a set of well-integrated cross-layer design techniques for interference management and system optimization. This proposed approach may considerably improve the rate performance and user capacity, and is promising in bridging the gap between theory and practice for broadband wireless access. To support the drastically increased mobile data traffic in wireless broadband services, this work focuses on a systematic cross-layer system optimization approach that relies on three major pillars: 1) at the physical layer, base stations with massive multiple-input multiple-output antenna systems are used; 2) at the wireless network architecture level, a multi-tier heterogeneous network approach is selected, achieving unprecedented spatial spectrum reuse; 3) at the cross-layer optimization level, a holistic network utility maximization approach is proposed, that systematically obtains layered protocol architectures from the structure of the global optimization solution. In relation to the above pillars, the fundamental challenges that will be addressed in this project are: 1) the design of integrated and power-efficient reconfigurable massive multiple-input multiple-output front-end antenna systems based on the concept of hybrid beamforming, i.e., on the optimal splitting of multiuser precoding and inter-cell interference management functions between digital baseband processing and analog radio frequency beamforming; 2) the design of hybrid beamforming schemes that exploit long-term channel statistics for inter-cell coordinated interference management, and instantaneous channel state information to achieve spatial multiplexing gain in each cell; 3) a user partitioning and scheduling approach based on clustering the user space according to quality of experience requirements, channel statistics and mobility, assigning network utility functions to the different user groups, solving the combined network utility maximization problem and systematically deriving a layered protocol architecture from the structural properties of the optimization solution. In addition, the work will significantly extend current mathematical performance analysis of wireless networks, based on advanced tools from stochastic geometry and random matrix theory, in order to assess quantitatively the performance gains over current technology of the proposed approach. Last, small-scale experiments will be conducted with software-defined radios equipped with the front end that will be developed.