The abundance of spectrum in mmwave provides a way forward to satisfy the growing demand of higher throughput. However, the use of mmwave for wireless communicaton faces challenges due to its propagation characteristics, namely, line of sight (LoS) and large path loss. They limit the coverage of mmwave base stations and requires substantial overhead in beam scanning and alignment. We have invented smart mmwave antennas that overcome these challenges, making the wireless channel intelligent, active and controllable, revolutionizing the long-held concept that wireless channels can only by passive and lossy. More details will provided when we introduce our smart mmwave antenna products to the market.
Hybrid Beamforming (HB) has been widely studied for reducing the number of costly RF chains in massive MIMO systems. However, previous work on HB is limited to single UE or single group of UEs, using a frequency-flat first-level Analog Beamforming (AB) that cannot be applied to multiple groups of UEs served in different frequency resources in an OFDM system. We have developed a novel HB technology with unified AB based on the spatial covariance matrix of all UEs is proposed for a massive MIMO-OFDM system to support multiple groups of UEs.
A massive MIMO base station embodying our HB technology can use a much smaller number of RF chains to serve multiple groups of UEs and achieve more than 95% performance of full digital beamforming, thus offering significant cost and power savings without large loss in performance. This is especially important for massive MIMO in FDD in the sub-6GHz bands. Many FDD bands are widely deployed around the world, but to date massive MIMO development has been limited to TDD only. Our HB technology brings the power of massive MIMO to the FDD bands.
Despite the theoretical benefits, in practice, implementing a large number of Radio Frequency (RF) chains in a massive MIMO system can be challenging, since it increases the system cost and power consumption, and lowers power efficiency. To address these issues, Hybrid Beamforming has been proposed, where a high-dimensional first-level RF Analog Beamforming (AB), which could be realized by a low-cost Phase Shift Network (PSN), is applied to decrease the number of RF chains before the reduced-dimensional second-level digital baseband beamforming is employed. However, previous work on HB suffers from the following limitations.
1) In sub 6GHz bands, prior art HB methods proposed for Single-User MIMO (SU-MIMO) systems and for MU-MIMO systems are frequency-flat AB methods, designed for a single UE or a single group of UEs, where all UEs in the group are served on the same frequency resource. These methods cannot be applied to multiple groups of UEs with each group being served by a different frequency resource. A massive MIMO HB method supporting multiple groups of UEs offers advantages over methods that only support single UE or single group of UEs.
Prior art HB methods that allocate the whole band to a single UE or a single group of UEs suffer from low flexibility in UE scheduling, which may cause high latency due to the relatively small number of scheduled UEs per scheduling period, and low resource use efficiency because some UEs do not need to be served by the whole band.
2) Even in the case that the whole frequency band is allocated for a single UE or a single group of UEs, challenges still exist to apply prior art HB algorithms in massive MIMO-OFDM systems. Specifically, full Channel State Information (CSI) is assumed at either each BS or each UE, which is challenging to be acquired with HB because the number of RF chains is much smaller than the number of antennas. In the ideal case where each RF chain is connected to all antennas, although it is possible to estimate full CSI in succession, it would significantly increase the channel estimation overhead and system complexity, which are already causing difficulties in realizing aggressive spatial multiplexing. In addition, in the more practical case where each RF chain is only connected to a subset of antennas, it is not even possible to acquire full CSI.
3) Prior art HB algorithms for SU-MIMO systems and for MU-MIMO systems are specifically designed for mmWave systems where the number of multipath components are very limited and Line-of-Sight (LoS) channels are generally assumed.
They cannot be directly applied to sub 6GHz frequency bands where many multipath components exist and Non- LoS (NLoS) channels are very common.
Our novel HB technology with unified AB that overcomes the limitations identified above. It is specifically designed for massive MIMO-OFDM systems in sub 6GHz frequency bands to support MU-MIMO beamforming to multiple groups of UEs. Instead of full CSI assumed by prior HB methods, our HB technology is based on the second-order spatial covariance matrix (SCM), which is a much more practical assumption than full CSI. Simulation results show that our HB technology with reduced number of RF chains can achieve no less than 95% performance of complete digital beamforming with much higher number of RF chains.