Paolo Baracca (Nokia Bell Labs)
Exploiting high carrier frequencies for mobile communications is a fundamental enabler to cope with the ever-increasing throughput demand. In fact, millimeter wave (mmWave) bands offer huge portions of spectrum that can be used to deliver very high data rates from or toward the mobile users. When compared to current 4G systems working at frequencies below 6 GHz, mmWave base stations and users are expected to be equipped with many more antennas, which is viable due to the reduced antenna element size. Therefore, multiple-input multiple-output (MIMO) precoding in the form of either beamforming or spatial multiplexing will be adopted. However, which MIMO scheme, at these high carrier frequencies, represents the best tradeoff in terms of energy consumption, cost and performance, is still under investigation. As explained in [1, Chapter 6], there are three main MIMO architecture alternatives for mmWave systems: analog, fully-digital and hybrid beamforming.
Analog beamforming
Users operating at mmWave frequencies are usually more noise- than interference-limited and, therefore, good performance can be obtained by applying a simple beamforming technique where: users are multiplexed in the time domain, only one data stream per time slot is sent and all the antennas are used to provide array gain to compensate the high path-loss characterizing mmWave bands [2]. As illustrated in Figure 1(a), this scheme can be realized, for instance at the transmitter side, by using a fully analog beamforming architecture, where only one digital-to-analog converter (DAC) and one radio frequency (RF) chain are required. Namely, analog beamforming is performed in the RF analog domain, for instance, by using M phase shifters, each one mapped to an antenna element. This architecture allows the transmitter to generate a wideband beam that focuses the power toward a specific direction to increase the signal to noise ratio (SNR) at the receiver. Nevertheless, a fully analog beamforming architecture has also some limitations such as the fact that beams are wideband and spatial multiplexing is not allowed.
Fully-digital beamforming
System performance can strongly improve by using a fully-digital beamforming architecture, as shown in Figure 1(b), with one RF chain per antenna element. By implementing the precoder in the digital baseband (BB) domain, this transceiver allows to both a) implement different precoders on different sub-bands (with the aim of compensating the frequency selectivity of the channel) and b) perform multi-stream transmission, for instance to simultaneously serve two line-of-sight users that are physically separated. Due to the large number of RF chains required, the main drawbacks of this architecture are the high cost and energy consumption. More effort is needed to make this architecture feasible at mmWave. However, some preliminary studies have already shown interesting results into this direction, for example by using very low resolution analog-to-digital converters to strongly decrease the cost/energy consumption of the transceiver [3].
Hybrid beamforming
A compromise between the two architectures described above is the hybrid beamforming architecture shown in Figure 1(c). Although different configurations are possible [4, Figure 2], the general idea consists in equipping the device with P RF chains, being P much smaller than the number of antennas, i.e., P<<M, thus still allowing some precoding flexibility but at a reduced cost and power consumption. Hybrid beamforming architectures can be used in different complementary ways. For instance, multiple beams in the analog domain can be used to send a data stream to a specific user whose channel has few strong paths, i.e., the multiple beams are used to compensate the multipath fading and, in turn, to increase the SNR. As an alternative, spatial multiplexing to serve multiple users can be implemented with a joint design of a) wideband beams in the RF analog domain and b) some more advanced per-sub-band precoding, like zero-forcing, in the digital BB domain. Several works have already shown that hybrid beamforming can achieve the performance of fully-digital beamforming in scenarios with one base station serving one or multiple users. However, the gap among these two architectures tends to increase when RF impairments are included, showing that the performance of hybrid beamforming can significantly vary depending on the specific hardware implementation [5]. Recently, we have performed a rather comprehensive system level analysis comparing these beamforming options with many interfering base stations that are serving users in different channel conditions, thus requiring also some user selection algorithms [6, 7]. Our results, targeting dense urban scenarios and taking also hardware impairments into account, have confirmed that hybrid beamforming represents a good tradeoff for mmWave mobile communications, being able to obtain performance close to fully-digital beamforming in many relevant cases.

Figure 1: MIMO architectures for mmWave mobile transmission: (a) analog beamforming, (b) fully-digital beamforming and (c) hybrid beamforming.
References
[1] A. Osseiran, J. F. Monserrat and P. Marsch, “5G mobile and wireless communications technology,” Cambridge University Press, 2016.
[2] S. Sun, T. S. Rappaport, R. W. Heath, A. Nix and S. Rangan, “MIMO for millimeter-wave wireless communications: beamforming, spatial multiplexing, or both?,” in IEEE Communications Magazine, vol. 52, no. 12, pp. 110-121, Dec. 2014.
[3] R. W. Heath, N. González-Prelcic, S. Rangan, W. Roh and A. M. Sayeed, “An overview of signal processing techniques for millimeter wave MIMO systems,” in IEEE Journal of Selected Topics in Signal Processing, vol. 10, no. 3, pp. 436-453, Apr. 2016.
[4] R. Méndez-Rial, C. Rusu, N. González-Prelcic, A. Alkhateeb and R. W. Heath, “Hybrid MIMO architectures for millimeter wave communications: phase shifters or switches?,” in IEEE Access, vol. 4, pp. 247-267, Jan. 2016.
[5] A. Garcia-Rodriguez, V. Venkateswaran, P. Rulikowski and C. Masouros, “Hybrid analog–digital precoding revisited under realistic RF modeling,” in IEEE Wireless Communications Letters, vol. 5, no. 5, pp. 528-531, Oct. 2016.
[6] S. Gimenez, S. Roger, D. Martín-Sacristán, J. F. Monserrat, P. Baracca, V. Braun and H. Halbauer, “Performance of hybrid beamforming for mmW multi-antenna systems in dense urban scenarios,” in Proc. IEEE International Symposium on Personal, Indoor and Mobile Radio Communications (PIMRC), Valencia (Spain), Sep. 2016.
[7] S. Gimenez, S. Roger, P. Baracca, D. Martín-Sacristán, J. F. Monserrat, V. Braun and H. Halbauer, “Performance evaluation of analog beamforming with hardware impairments for mmW massive MIMO communication in an urban scenario,” MDPI Sensors, 16(10):1555, Sep. 2016, doi: 10.3390/s16101555.