基於壓縮通道感知之前瞻波束成形於毫米波大型天線陣列系統

Advanced Beamforming based on Compressed Channel Sensing for Millimeter-Wave Large-Scale Antenna Systems

蔡承融(Cheng-Rung Tsai)

指導教授 ： 吳安宇

國立臺灣大學/電機資訊學院/電子工程學研究所/博士(2018年)

https://doi.org/10.6342/NTU201800727

摘要

國際行動通訊組織(International Mobile Communications, IMT)於IMT-2020為第五代行動網路(5th-generation Mobile Networks, 5G)制定了未來2020年行動通訊發展的框架和總體目標。其中5G的發展目標除了要增強過去的行動寬頻業務(Enhanced Mobile Broadband, eMBB)以提供更高的資料傳輸速率與更大的系統傳輸容量，還需要滿足未來更多樣化的使用場景需求。例如在物聯網(Internet of Things, IoT)應用場景中提供大量終端裝置聯網的需求(Massive Machine Type Communication, mMTC)，或是在車聯網(Vehicle to Everything, V2X)應用場景中提供超高可靠度低延遲通訊(Ultra-Reliable & Low-Latency Communication, uRLLC)。為了同時滿足多樣性的需求，目前第三代合作夥伴計畫(3rd generation partnership project, 3GPP)在針對5G所制定的新式無線電(New Radio, NR)標準中朝向毫米波(millimeter-wave, mmWave)頻段尋求更大的頻譜資源，並且提供更有彈性的資源調度能力。此外為了避免毫米波信號傳輸所產生的能量大幅衰減，使用大型天線陣列系統(large-scale antenna array system, LSAS)實現波束成形(beamforming)增益，克服傳輸能量衰減以確保訊號的傳輸品質也成為5G NR必要的技術發展項目。

然而，過去的無線通訊系統(如4G LTE與Wi-Fi)因為天線單元數量有限，在處理多天線訊號可以皆在數位基頻進行。在毫米波大型天線系統中，天線單元數量可以輕易地達到上百個！為了提供每一個天線單元獨立的基頻/射頻訊號轉換鏈路(baseband/RF converting chain, RF chain)，大型天線陣列勢必會產生巨大的功率消耗以及硬體成本。為了避免使用大量的基頻/射頻訊號轉換鏈路，採用混合數位-類比大型天線陣列系統(Hybrid Analog-Digital LSAS)目前已經為高通、三星、華為等領導廠商之主要研發方向。除此之外，由於毫米波大型天線陣列使用更大的頻寬以及更多的天線，通道狀態資訊(channel-state-information, CSI)量也會大幅成長。基地台與手機端為了取得通道狀態資訊所需要的訓練或資源負擔與運算複雜度都會遽增。如何利用毫米波傳輸通道的特色達到低資源負擔且具有效性的通道狀態資訊取得與使用方式，成為毫米波大型天線系統最重要的議題。

為了克服上述挑戰，本論文率先引進於近年新興的壓縮感知(compressed sensing)訊號處理技術，利用毫米波通道的稀疏特性(sparse nature)，提出壓縮通道感知(compressed channel sensing)技術。此壓縮通道感知技術可大幅降低基地台與手機端為了取得通道狀態資訊所需要的訓練或資源負擔，並維持低信令開銷(signaling overhead)與估測強健性(robustness)。本論文也提出基於多測量向量(multiple measurement vectors, MMV)之毫米波通道估測演算法，大幅降低在進行壓縮通道估測時的運算複雜度。另一方面，此估測演算法能夠在毫米波通道具有結構化稀疏時仍只需使用少量的參考訊號資源。基於壓縮感知所取得的低維度通道狀態資訊，本論文也出波束空間混合數位-類比波束成型演算法(beamspace hybrid analog-digital beamforming algorithm)。在稀疏毫米波通道之下，基於波束空間通道狀態資訊的通道維度可以遠小於原始空間通道資訊的通道維度，因此在設計混合數位-類比波束成型時能夠省下大量運算複雜度。。除此之外，本論文更進一步將此壓縮通道感知技術拓展至5G NR系統。最後，本論文使用了3GPP所制定之通道模型，對於本論文所提出之壓縮通道感知技術進行連結級的驗證(link level evaluation)。

此論文針對於毫米波大型天線陣列系統提出多項基於壓縮感知的前瞻性波束成形技術，並希望能夠成為下世代行動通訊的關鍵技術。

關鍵字

毫米波通訊 ； 大型天線陣列 ； 壓縮通道感知 ； 壓縮通道估測

並列摘要

In IMT-2020, 5G will bring a whole lot more by supplementing on enhanced Mobile Broadband (eMBB) and including some radically new capabilities, such as massive Machine Type Communication (mMTC) and ultra-Reliable and Low-Latency Communication (uRLLC). To meet the diversity to serve a wide range of applications, the new radio interface (NR) technical specifications for 5G laid by 3rd Generation Partnership Project (3GPP) focus on millimeter-wave (mmWave) bands to utilize the abundant spectrum resource, and aim to provide flexible frameworks and self-contained designs to allow network operators to more efficient multiplex diverse services on a unified 5G network, while also ensuring 5G NR forward compatibility to future 5G features and services. On the other hand, large-scale antenna system (LSAS) is considered as an essential technology in 5G NR to realize beamforming gain to compensate for the huge propagation loss in mmWave communications.

Millimeter-wave LASA requires hundreds of antenna elements to realize sufficient beamforming gain, equipping every antenna with an individual radio frequency (RF) chain along with high-frequency mixed-signal components would incur high hardware cost, complexity and power consumption, particularly in the context of consumer electronics. To avoid the massive number of RF chains, hybrid beamforming is widely adopted for mmWave LASAs. Moreover, since the mmWave LSAS exploits much larger bandwidth and much more antenna elements, quantity of channel-state- information (CSI) grows exponentially. To acquire CSI for mmWave LSASs, the resource/training overhead and computational complexity are increased prohibitively. Therefore, how to exploit the characteristics of mmWave channels to achieve low training/resource overhead, and how to efficiently acquire and utilize the CSI, have become one of the most important issues in mmWave LSASs.

To overcome the aforementioned challenges, this dissertation proposes compressed channel sensing (CCS) based on compressed sensing (CS) to exploit the sparse nature in mmWave channels. The proposed CCS can to reduce training/resource overhead for acquiring CSI from mmWave LSASs, while maintaining low signaling overhead and robust channel estimation. We also propose a low-complexity compressed channel estimation to acquire CSI from the measurement of CCS. Moreover, when the mmWave channels exhibit structured sparsity, the proposed estimation still requires only small resource/training overhead. Then, we develop a beamspace hybrid beamforming algorithm based on the beamspace CSI acquired by the proposed compressed channel estimation. In mmWave LSAS, due to the sparse nature, the dimension of beamspace channel is much smaller than that of the original spatial channel. Thus, the proposed beamspace hybrid beamforming can greatly reduce the computational complexity. We further extend the proposed CCS framework for 5G NR systems. Finally, to show the feasibility and superiority, we adopt the 3GPP channel model to evaluate the techniques proposed in this dissertation in link level.

In summary, this dissertation presents several advanced beamforming techniques based on compressed channel sensing aimed to be the key enabling techniques for mmWave communications.

並列關鍵字

Millimeter-wave communications ； Large-scale antenna systems ； Compressed channel sensing ； Compressed channel estimation

參考文獻

[1] The 5G Infrastructure Public Private Partnership (5G-PPP) Technical Report, February 2016, 5G empowering vertical industries.

[2] 3GPP Technical Report TR38.913, Release 14, March 2017, Study on scenarios and requirements for next generation access technologies.

[3] L. Wei, R. Q. Hu, Y. Qian, and G. Wu, “Key elements to enable millimeter wave communications for 5G wireless systems,” IEEE Wireless Commun., vol. 21, no. 6, pp. 136–143, Dec. 2014.

[4] W. Roh, J.-Y. Seol, J. Park, B. Lee, J. Lee, Y. Kim, J. Cho, K. Cheun, and F. Aryanfar, “Millimeter-wave beamforming as an enabling technology for 5G cellular communications: theoretical feasibility and prototype results,” IEEE Commun. Mag., vol. 52, no. 2, pp. 106–113, 2014.

[5] X. Zhang, A. F. Molisch, and S. Kung, “Variable-phase-shift-based RF-baseband codesign for MIMO antenna selection,” IEEE Trans. Signal Process., vol. 53, no. 11, pp. 4091–4103, Nov. 2005.