The Transformative Potential of Massive MIMO

Subscribe To Download This Insight

1Q 2019 | IN-5420

Our increasingly mobile-centric world requires an increased focus on antenna improvements and deployments in large numbers. Massive MIMO antennas offer the advantages of extended ranges, increased data throughputs, and more thorough coverage, but the challenges will require extensive R&D.

Registered users can unlock up to five pieces of premium content each month.

Log in or register to unlock this Insight.


The Promises of Massive MIMO


As mobile networks transition to 5G, massive Multiple Input, Multiple Output (MIMO) antennas will become essential in helping to realize the predicted 100X capacity improvement when compared to today’s Long-Term Evolution-Advanced (LTE-A) and LTE-A Pro technologies. Massive MIMO antennas will be required in the low-, mid-, and high-band spectrums. Massive MIMO antennas are an essential part of the toolbox for 5G, but they are also increasingly important in the LTE-Advanced Pro/Gigabit LTE Radio Access Networks (RANs).

A massive MIMO antenna deploys large numbers of antenna elements inside the antenna enclosure—many more than there are simultaneous active subscribers in the sector. These “excess” antenna elements permit the RAN to narrowly focus the Radio Frequency (RF) energy only toward those subscribers that require it. The effect of this beamforming is to dramatically increase the sector spectral efficiency, allowing the sector throughput to increase and the Mobile Network Operator (MNO) to maximize the use of network resources in the RAN. Also, because the RF energy is focused into highly directive narrow beams, there is very little interference leakage between beams of neighboring cells, and the RAN can achieve higher levels of densification and more simultaneous users.

High antenna element count massive MIMO systems can also leverage the benefits from channel hardening and link reliability. Variations in the radio channel caused by multipath or reflected components reduce as the number of antenna elements increases, and the narrow beams operate in favorable radio propagation conditions and are orthogonal to each other. Channel hardening averages out the noise, fast fading, and hardware impairments in the channel. There is reduced variability in the radio channel, so fewer retransmissions are required, resulting in more predictable performance due to the decreasing likelihood of lost packets. These attributes are important for Ultra Reliable Low Latency Communications (URLLC) and massive Machine Type Communications (mMTC) use cases in 5G.

The massive MIMO array gain, which becomes higher as the number of antenna elements increases, can also be leveraged to give a range extension and increase data throughput rates in locations already covered, or provide coverage in previously unavailable locations, such as indoors. Alternatively, a high array gain can be used to lower the transmit power in the downlink, simplifying and reducing the cost of the hardware. As a result, massive MIMO systems can be designed to use inexpensive low-power components, because the total radiated power in the sector is now shared among many antenna elements. This is especially important in Internet of things (IoT) applications with sensors or actuators required to have battery life of 10 years or more, where uplink transmission may only require low order binary modulation. As a result, massive MIMO enables a reduction in the total RF power of the sector leading to Total Cost of Ownership (TCO) savings for the MNO and potentially raising the prospect of solar or wind-powered basestations.

The Problems with Massive MIMO


While massive MIMO technology offers these advantages, it does not come without its challenges. Several of these are discussed below and ABI Research anticipates that moving massive MIMO deployments from the test and pilot stage to commercial at-scale operations in live networks may resolve many of the challenges.

A massive MIMO system relies on obtaining good Channel State Information (CSI), so that appropriate beamforming and beam tracking signal processing can occur. The conventional grid-of-beams arrangement where typically eight beams are used in the downlink and the user terminal reports the best channel using the uplink is simple and works for both Time Division Duplex (TDD) and Frequency Division Duplex (FDD) access schemes. However, the CSI estimation using this framework is never a perfect match, yielding only approximate results and diminishing the performance gains from the massive MIMO RAN. A much better arrangement is when the user transmits a pilot signal in the uplink and the basestation estimates the CSI. This provides accurate CSI to the system for the best massive MIMO performance. However, this only works for TDD when the uplink CSI can be used for the downlink, because the channels are the same or reciprocal. For FDD systems, where the uplink and downlink channels are different, implementing high-performance massive MIMO systems remains a challenge, although many vendors are commercializing FDD massive MIMO systems now.

Pilot contamination is another challenge in the massive MIMO RAN. This occurs because the maximum number of orthogonal pilot sequences can become exhausted using the RAN. The re-use of pilot sequences from one cell to another is termed “pilot contamination.” This results in received channel estimates at the basestation being contaminated by a combination of estimates from the pilots in other channels that share the same pilot sequence. Pilot contamination can be dealt with in several ways. The allocation of pilot signals can be done in a less aggressive way, so that mutually contaminating cells are separated in frequency, or by coordinating the use of pilots or adaptively allocating pilot sequences to different terminals. The optimum strategy for minimizing pilot contamination is an area for active research.

A digital beamforming architecture for massive MIMO promises the best performance in theory, because each individual antenna element can receive the optimum signal amplitude and phase. However, a digital beamforming architecture is not practical or economic for high-count antenna arrays, and without a breakthrough in digital beamforming architectures, a hybrid architecture remains the architecture of choice for very high-count arrays. Beamforming architecture requires further research to develop algorithms, characterize performance, and optimize energy use to approach parity with the ideal digital beamforming architectures and reach its full potential.

Massive MIMO for high-mobility or fast-moving terminals is another active research topic. There is an upper boundary on the speed of a moving terminal that is inversely proportional to the carrier frequency, and at high frequencies, a terminal could move faster than the CSI could be updated. Also, high mobility presents a challenge to TDD systems more than FDD systems, because uplink and downlink transmissions occur simultaneously in FDD. Additional research can help determine the best way to support robust high-mobility terminals at high carrier frequencies.

While massive MIMO increases sector capacity, in some situations, alternative solutions can be more effective, such as when users are not sufficiently separated or are clustered together. In these situations, the use of small cells or sectorization may be more effective for handling traffic at the hotspot. Another drawback of massive MIMO is high-power consumption due to the large amount of digital signal processing occurring in the baseband and basestation. Hence, depending on the mobile environment and deployment scenario, small cell antenna solutions may offer better results than massive MIMO.

While this is a partial list of the challenges inherent in massive MIMO, ABI Research’s report The Rise of Massive MIMO (AN-2758) offers a more in-depth discussion.

The Future of Massive MIMO


The promises of massive MIMO are so compelling that a very rich Research and Development (R&D) ecosystem has arisen, as the vendors drive to solve the challenges ABI Research describes.

The rich field of R&D topics include algorithms for power-efficient and economic signal processing. Massive MIMO antennas generate enormous amounts of baseband data that need to be processed in real time, and consequently, the signal processing needs to be simple. Research is required on the design and implementation of optimized signal processing algorithms, which lowers the power and cost of the massive MIMO array.

Optimum hardware architectures for efficient low-power signal processing is another research topic that is required so that the ideal massive MIMO digital architecture can become a practical reality. While massive MIMO can be implemented with low-cost hardware, the manufacture of hundreds of RF chains, Digital to Analog Converters (DACs), Analog to Digital Converters (ADCs), and up/down converters will require manufacturing economies of scale like those seen for the manufacture of mobile handsets.

Meta-materialspromise to drastically reduce the power consumption and cost of implementing a phased array and RF chain. One meta-material-based technique is called holographic beamforming, which implements phase changes at each antenna element with the appropriate reverse bias of a varactor diode associated with each antenna.

It is not yet clear which antenna patterns provide the best massive MIMO performance, although most pre-commercial antennas are arranged in two-dimensional planar arrays. Alternative antenna geometry, including cylindrical or curved arrays, may lead to significant performance improvements according to the deployment scenario and sector geometry. Dense urban areas with many high-rise buildings may require beamforming that has a wide angular range in the vertical dimension. This contrasts with the antenna pattern that would be most appropriate for suburban or rural deployments where the horizontal angular range should be large and where the vertical range may not even be required. These considerations may even lead to the concept of massive MIMO antenna patterns that are specific to regions, because dense urban buildings in many Asian cities are very high compared to European dense urban settings where buildings do not rise as high vertically, for example.

Other R&D topics are distributed massive MIMO or 5G Coordinated Multi-Point (CoMP), the use of Artificial Intelligence/Machine Learning (AI/ML) for beamforming, and beam tracking and the use of alternative structures for the antenna based on waveguide or other technologies.

While this is not an exhaustive list, ABI Research anticipates that many of these challenges will be solved as massive MIMO becomes mainstream and 5G rolls out over the next 10 years. As the last remaining physical link between the network and the radio channel, the antenna hardware cannot be virtualized, and the successful massive MIMO vendor will innovate to solve these challenges and monetize them for commercial success.