5G cellular systems are transforming not only the way that we are communicating, but also enable a host of new applications that ? while less visible to consumers ? are equally groundbreaking. It has been widely recognized that 5G communications is an essential part of modern infrastructure and basis for economic development in the US. While 5G standardization and deployment already started in the late 2010s, it is an ongoing process, and modifications and improvements will continue to be provided far into the 2020s.

An essential part of 5G is the use of the millimeter-wave spectrum. This frequency range offers much larger bandwidth, and thus higher data rates, than the frequency bands that have traditionally been used for cellular systems and WiFi. Consequently, investigations of communications systems operating in the millimeter-wave band are of pivotal importance. This project, which lasted from 2016 to 2021, explored two key aspects of millimeter-wave systems: beamforming, and channel characterization.

A key challenge of operating in the mmWave frequency range is the increased attenuation, which results in small coverage range of a base station ? in other words, a user has to be very close to a base station in order to communicate. This problem can be solved, at least partly, with the use of antenna arrays at both base station and user handset. In the simplest form, at the base station, the beams point towards the user, and at the user handset, the beams point towards the base station. However, conventional antenna arrays need full radio chains for each antenna element, which is both expensive and energy-hungry. A more cost- and energy-efficient way is hybrid digital/analog (HDA) architecture, which consists of two concatenated beamformers in the analog and digital domain, respectively. An important part of this project was the investigation of such HDA structures, and algorithms to adapt their settings to the particular channel conditions between base station and user handsets. We in particular considered situations where the analog beamformer is adapted based on average channel characteristics (averaged over quick variations/fades), while the digital beamformer follows the instantaneous variations. This approach has multiple advantages, including reduced overhead for pilot tones, greater robustness, and reduced requirements for the hardware used in the analog beamformers. During the course of this project, we have developed a variety of algorithms that jointly optimize the analog and digital beamformers at both base station and user handset, and furthermore take into account that a base station can talk to multiple users simultaneously, on the same frequency, to increase the overall throughput of the network. These results were published in journal papers [1-3], as well as multiple conference papers.

A further structure of interest is the combination of HDA with selection techniques, such that out of the many possible outputs of the RF beamformer, only a subset is converted to baseband and further processed. Such techniques are energy-efficient and can be well-implemented in practice, though their theoretical analysis is very complex. In [4] we derived a fundamental mathematical methodology, and applied it to the optimization and analysis of such systems. We also developed a new channel estimator that requires less hardware [5-6].

Besides all this work on 5G transceivers, we also performed detailed investigations of mmWave propagation channels, as they ultimately determine the performance of the algorithms and their properties should be taken into account for system optimization. Our work here ranged from the development of measurement instrumentation (channel sounders), to performance of measurement campaigns, to calibration procedures and signal processing for high-resolution evaluation of the measurements, to analysis of measurement campaigns and creation of channel models. For example, we developed a technique that allows the extraction of the fading depth (more precisely, the so-called Rice factor) from just a single measurement of the propagation channel.  Multiple journal papers (not listed here for space reasons) report our results.

[1] Li, Zheda and Han, Shengqian and Molisch, Andreas F (2017). Optimizing channel-statistics-based analog beamforming for millimeter-wave multiuser massive MIMO downlink. IEEE TransWirelessComm. 16 (7), 4288-4303.

[2] Li, Zheda and Han, Shengqian and Molisch, Andreas F (2018). User-centric virtual sectorization for millimeter-wave massive MIMO downlink. IEEE TransWirelessComm. 17 (1), 445460.

[3] Li, Zheda and Han, Shengqian and Sangodoyin, Seun and Wang, Rui and Molisch, Andreas F (2018). Joint Optimization of Hybrid Beamforming for Multi-User Massive MIMO Downlink. TransWirelessComm. 17 (6), 3600-3614.

[4] Ratnam, V.V. and Molisch, Andreas F.  and Bursalioglu, O. Y. and H. C. {Papadopoulos} (2018). Hybrid Beamforming With Selection for Multiuser Massive MIMO Systems. IEEE TransSignalProcess. 66 (15), 4105-4120.

[5] Ratnam, V.V. and Molisch, Andreas F.  (2019). Periodic Analog Channel Estimation Aided Beamforming for Massive MIMO Systems. IEEE TransWirelessComm. 18 (3), 1581-1594.

[6] Ratnam, V.V. and Molisch, A.F. (2019). Continuous Analog Channel Estimation-Aided Beamforming for Massive MIMO Systems. IEEE TransWirelessComm. 18 (12), 5557-5570.

Last Modified: 11/30/2021
Modified by: Andreas Molisch