Massive MIMO (mMIMO) is all about adding radio frequency (RF) chains, while not necessarily adding more radiating elements. Massive refers to the larger number of transceivers in the base station antenna array. The greater the number of transceivers that the antenna is equipped with, the more the degrees of freedom to modify the radiation pattern of the transmitted signal based on where the receiver is located. Another important feature is multiple RF chains so that several users can be spatially multiplexed when there are multiple antennas. For this, at least as many RF inputs are needed as users. Each can get the full array gain and the digital precoding can be used to avoid inter-user interference. This results in higher beamforming granularity and potential to utilise more degrees of freedom for more simultaneous layers and MU-MIMO as long as the signal quality and power levels are acceptable. In mMIMO, the number of degrees of freedom is usually much larger than the number of layers to be transmitted. With large antenna arrays, conventional signal processing techniques like maximum likelihood detection become prohibitively complex. The main question is whether huge multiplexing gains can still be achieved with low complexity signal processing and low-cost hardware implementation.
mMIMO in TDD
To take advantage of multipath, the spatial channel between antenna elements in the base station and user equipment (UE) is characterised by channel state information (CSI), which must be measured. The CSI is used to digitally encode and decode the data. In time division duplex (TDD) systems, where uplink (UL) and downlink (DL) channels are the same, downlink and uplink channels are said to be reciprocal. Having reciprocity means the channel needs to be measured only in one direction where the CSI can be directly retrieved from a reference signal for the purpose of channel characterisation. The UL CSI can be measured on the received sounding reference signal (SRS) transmitted from each UE and each UE’s SRS is received by all the antenna elements at the base station. In addition, all the computation intensive channel estimation and signal processing is performed at the base station. For this reason, an increase in the number of antenna ports does not lead to a proportionate rise in overheads in TDD systems. Care must be taken to ensure that the internal RF paths are properly calibrated to allow for the use of reciprocity.
In TDD, the overhead associated with UL channel estimation is proportional to the number of UEs, but is independent of the number of transmit antenna RF chains, thereby making the protocol fully scalable with respect to the degrees of freedom of atomic absorption spectroscopy. The overhead increases linearly with the number of active users in a cell. The derived CSI is then utilised in the DL due to the reciprocity between DL and UL radio channels. The discussion of reciprocity-based measurements is based on the reception of reference signals from UEs and is predicated on an acceptable signal quality level to perform CSI. If the quality on the received sounding signals is not acceptable, the CSI can be received as feedback from the respective UE performing measurements on reference signals in the DL (download link).
mMIMO in FDD
Most networks today in sub-6 GHz bands are operating in the frequency division duplex (FDD) mode, where the uplink and downlink use different frequency bands and channel reciprocity cannot be utilised. As a result, the FDD mode requires additional measurement reports to estimate the CSI for both bands. Due to the additional overhead, this need for dual-band CSI creates a performance penalty. For FDD systems, the downlink overhead incurred by reference signals for the CSI increases linearly with the number of transmitting base station antenna RF chains. Given overhead dependencies on both number of antenna ports and users in FDD, there are relatively more constraints and additional overheads compared to reciprocity-based direct channel measurements. A high number of DL transmitter antenna RF chains typically require a high number of orthogonal CSI reference signals in FDD. They would also require a higher UL overhead for reporting the estimated CSI from UEs to the base station. There will be a balance of CSI-reference signal overhead versus user data transmission within a transmission block. This trade-off is between reporting of more relevant channel components with a higher accuracy versus reducing the number of resource elements available for data transmission, thus limiting the maximum data throughput. Currently, commercial mMIMO systems in FDD rely on 16 or 32 transceivers design.
For both FDD and TDD, the number of sub-arrays in architecture defines how well the mMIMO system will work in a given environment.
(Based on an extract from 5G Americas’ white paper, “Advanced Antenna Systems for 5G White Paper”.)