Robust Algorithm Development and Demonstration for Agile mmWave MIMO Hybrid Beamforming

毫米波（mmWave）频带的通信有望实现移动的数据业务的快速增长。然而，利用它们的主要限制是它们的高路径损耗，需要由具有许多元件的天线阵列来补偿。反过来，需要使用波束成形技术来最佳地向用户分配信号能量。由于纯数字实现可能消耗太多功率，因此可以通过结合数字基带处理利用模拟波束成形（称为“混合”波束成形）来实现复杂度、硬件成本和性能之间的良好折衷。毫米波系统的相干时间预期比低频系统的相干时间短得多，低频系统不是由强各向异性波束图案服务，而是仅由弱各向异性波束图案服务。因此，在毫米波波段加速（模拟）波束选择的算法对于一般天线阵列的成功应用以及特别是对于使用混合波束成形方法将是至关重要的。在这个项目中，我们计划对混合波束形成系统的自适应和加速波束选择进行基础研究，以及研究硬件相关问题及其补偿。只有在此基础上，一个切实可行的制度实施才显得可行。与传统的基于固定长度测试的波束选择方法相比，采用可变长度训练序列的混合自适应算法可以获得更高的时间效率。我们已经证明了这一点的情况下，一个单一的用户，我们提出的“顺序竞争”的测试。这种鲁棒算法还可以作为构建块，将数字波束成形的标准自适应算法扩展到适合混合波束成形的自适应算法。除了算法之外，还需要表征硬件相关损伤对混合波束成形性能的影响，特别是对于宽带宽以及具有多个RF链的MIMO设置。这包括例如来自次优硬件组件的频率选择性、相位噪声等，如果不适当地考虑，则可能导致理论上证明的算法的降级或故障。在数字基带层面，需要开发能够同时处理多个数据流和用户数Gbit/s数据速率的高效信号处理方案，并需要一个合适的低延迟控制平面来协调数字和模拟波束形成部分。为了验证所开发的系统概念，将在接入场景中进行60 GHz硬件演示。模拟波束形成部分将使用项目合作伙伴提供的矢量调制器进行。这将为混合波束成形的可行性以及硬件约束对性能的影响提供有价值的见解。

Communication at millimeter wave (mmWave) bands promises to enable the rapid growth of mobile data traffic. However, the major limitation of utilizing them is their high path loss that needs to be compensated by antenna arrays with many elements. In turn beamforming techniques need to be used to optimally distribute the signal energy towards the users. As a purely digital implementation might consume too much power, a good trade-off between complexity, hardware cost, and performance can be achieved by utilizing analog beamforming in conjunction with digital baseband processing, known as ‘hybrid’ beamforming.Assuming high carrier frequencies together with user mobility occurring in the radio access network, the coherence time of mmWave systems is expected to be much shorter than that of low-frequency systems, that are served not by strongly but only weakly anisotropic beam patterns. Therefore, algorithms that accelerate (analog) beam selection at mmWave bands will be of vital importance for the successful application of antenna arrays in general and for using the hybrid beamforming approach in particular. In this project, we plan to do fundamental research on adaptive and accelerated beam selection for hybrid beamforming systems, as well as to investigate of hardware-related issues and their compen-sation. Only on this basis, a practical system implementation appears feasible. Compared to conven-tional approaches, like beam selection using fixed length tests, hybrid adaptive algorithms with variable length training sequences can achieve higher temporal efficiency. We have shown this already for the case of a single user by our proposed ‘sequential competition’ test. Such a robust algorithm can also serve as a building block to extend standard adaptive algorithms well-known for digital beamforming to adaptive algorithms suitable in a hybrid beamforming context.Apart from algorithms the impact of hardware-related impairments on the hybrid beamforming performance needs to be characterized, especially for wide bandwidths as well as for MIMO setups with multiple RF-chains. This includes e.g. frequency-selectivity, phase noise etc. from sub-optimal hardware components, that can lead to degradation or failure of the theoretically proven algorithms if not properly accounted for. On the digital baseband level, efficient signal processing schemes that can simultaneously handle multiple streams and users with data rates of several Gbit/s need to be developed and a suitable control plane with low latency is required to coordinate the digital and analog beamforming sections.For verification of the developed system concept, a 60 GHz hardware demonstration will be undertaken in an access scenario. The analog beamforming part will be performed using vector modulators provided by project partners. This will provide valuable insights into the feasibility of hybrid beamforming and the impact of hardware constraints on performance.