Understanding MIMO

MIMO (Multiple-In, Multiple-Out) is a radio access technology that uses multiple transmitters and multiple receivers to communicate multiple streams of data over the same frequency band simultaneously.

The demand for mobile data continues to explode, with an incredible projected year-on-year growth of 46% (Statista, 2019). Mobile network operators are continually pressed to find new ways to get more data through the air. Cellular-band radio frequency spectrum is an incredibly finite resource and accordingly buying exclusive operating rights commands a very high price. For example, in the US the price per 1 MHz of 4G / 5G suitable spectrum in 2017 pushed USD $240 million.

There are four approaches to increasing capacity without having to acquire more spectrum. All four circle around the idea of increasing something known as spectral efficiency. Spectral efficiency, measured in bits per second per Hertz (b/s/Hz), refers to how much information can be transmitted each second over a given width of radio spectrum. Limits to spectral efficiency are governed by the Shannon-Hartley Theorem.

Massive MIMO antenna internal view

Cell Densification

In even a small town there may be thousands of mobile data users, from ordinary pedestrians, to people driving in vehicles, to small IoT devices built into advertising signs and vending machines. While a single cell tower may be sufficient to cover this small town, doing so would put each device in competition with each other for airtime. The radio access network would have to then juggle a complicated schedule of who transmits when and how many resource blocks to dedicate to each user. Some users would be far from the cell site and have to transmit at a slower rate in order to make sure their information was sent successfully.

By adding additional cell towers in key geographic locations the total number of users will be divided between all sites. Each site now schedules a smaller number of users, meaning more airtime per user. This seems reasonably straight-forward, however as these new sites are operating on the same frequency band they must be carefully positioned and the transmit power of each base station coordinated to ensure interference is kept to a minimum.

Adding sites in weak signal areas allows users, who were originally transmitting at a slower rate and consuming more airtime per unit of data, to communicate with the base station at a faster rate and improving the number of bits per second.

Improve UE Signal Strengths

A common approach to increasing network capacity is employing the use of external antennas for stationary devices. This can come in a few forms, from connecting high gain antennas to user equipment sitting on the cell edge, to setting up the network to operate in a Fixed Wireless Access (FWA) mode. By correctly positioning each user's antenna the receive signal strength is increased in both directions. With a stronger and cleaner signal the device can communicate at a faster rate using a higher order modulation and coding scheme.

This improvement is not insignificant. The difference between the slowest and fastest (least complex and most complex) transmission schemes is anywhere from a 12 to 49-fold increase (depending on LTE/A/A-Pro/NR CQI mapping). This means up to a 49x increase in capacity is achievable simply by lifting users from low schemes to high schemes.

Upgrade to Higher Order Modulation

Without diving too deep into the topic, at a very basic level modulation represents the amount of information that can transmitted in a given unit of time. Modulation efficiency is measured in bits per symbol (b/symbol). In the context of digital transmission a 'symbol' refers to an identifiable pulse or tone, with the symbol rate representing the number of waveform changes occurring per time unit.

During the course of ordinary communication a 4G or 5G device will continually change its modulation scheme to reflect changes in the transmission conditions. If the transmission path becomes obstructed by objects, or the channel becomes noisy from interferers, the device will alert the base station which will look up a CQI Index to determine what modulation and coding scheme matches the newly reported signal quality.

Under clean conditions where signal strengths are good and channel noise is minimal the device will operate at the maximum modulation scheme available. As digital signal processing technology improves over time new modulation schemes become useable. Until recently (3GPP Rel. 12) the maximum downlink modulation scheme was 64QAM, meaning that even under perfect channel conditions devices were limited to 6 b/symbol. LTE-A Pro and 5G NR introduced 256QAM which provides 8 b/symbol resulting in an instant 33% increase in spectral efficiency for those devices operating under clean conditions.

A key advantage of this approach is that often the upgrade could come in the form of a software upgrade. The downside is that the new modulation scheme mostly affects those users operating under near perfect conditions. Of course with some users completing their data transmissions at faster rate the link utilisation is lower, freeing up scheduler resources that can be then dedicated to users suffering slower rates.

Multiple Antenna Techniques (MIMO)

This brings us to our final approach. MIMO involves multiple transmission antennas communicating with multiple receive antennas over the same frequency domain. Ordinarily two devices communicating on the one frequency causes a collision, or at the very least a rise in noise floor. MIMO uses the concept of "space" diversity to allow multiple transmissions over the one frequency without causing this negative impact. Spatial diversity involves physically separating antennas so that independent transmission paths can be established with the receiver.

The simplest form of antenna configuration is called SISO - Single In, Single Out. A device's RF front end is connected to a single antenna which establishes one communication link (referred to as a stream) to the receiving device's antenna. This forms a 1:1 relationship between the two devices antennas.

    MIMO - Multiple In, Multiple Out differs in that it uses more than one antenna at each end. Relationships can be one-to-many, or many-to-many. The most common form is 2x2 MIMO, which refers to two transmit antennas establishing two communication links (streams) with two receive antennas. It is important to note that the device may choose to not use all antennas, instead constructing fewer spatial streams depending on the transmission environment. This concept is important to understanding Massive MIMO.

    As a partial implementation, some technologies such as 3G use transmit or receive diversity which refers to using a single antenna at one end but multiple at the other. These configurations work best in multipath environments where the two antennas can detect and combine signals over different propagation paths.

    • Transmit Diversity: The use of two antennas at the base station which send identical streams of data to a single receiver antenna using Space Time Block Coding (STBC). The burden of complexity is borne by the base station instead of the user equipment.
    • Receive Diversity: The use of two antennas at the receiver which work in unison to help reconstruct the original base station transmission.
    SISO vs MIMO Channel Capacity Comparison [1]

    Using MIMO we can theoretically achieve up to an n increase in throughput per n antenna. Intuitively this follows from the ability to add one additional stream of data per additional antenna. Under real world conditions however this is rarely the case. To understand why we will look at the major MIMO configurations in the sections below.

    1x1 SISO, or just SISO, represents the simplest form of radio communication. SISO devices use a single antenna to establish a single stream of data with a single antenna connected to the receiving device.

    Due to the ease of constructing omnidirectional broadcast antennas and tolerance to rain SISO communication is usually vertically polarised. One notable exception is television broadcast which often favours horizontal polarised transmission. Point-to-point microwave and mmWave communication is still often SISO due to the increased challenges in manufacturing multi-polarised antennas - particularly those in the E-Band (60 to 80 GHz) and above.

    SISO was popular in early radio communications such as in 2G, and 3G networks. Some SCADA and UHF packet-radio still use SISO due to the low data rate requirements. SISO is also still very popular in the manufacturing of low cost IoT frontends, such as those in NB-IoT, LoRa, and so on.

    2x2 MIMO, sometimes referred to as 2T2R, uses two antennas to establish up to two streams of data with the receiving device. Compared to ordinary single antenna networks, 2x2 offers up to a 100% increase in throughput.

    With two spatial streams established, the data payload is divided across both antennas and transmitted over the same frequency band. In order for spatial multiplexing to be effective, the antennas must be well isolated and configured to provide a low correlation coefficient. Typically the most effective way to achieve low correlation in a 2x2 system is to use orthogonal polarisations, for example, using one vertically polarised antenna and one horizontally polarised antenna.

    Of course in many UE systems totally orthogonal polarisations are not possible, in which case the concept of Envelope Correlation Coefficient (ECC) comes into play. ECC accounts for characteristics such as 3D radiation patterns and phasing to determine how independent the two antennas are of each other. Ultimately practicality forms an upper limit to antenna performance as both antennas must be enclosed in a single device or installed in a small amount of space.

    2x2 MIMO block diagram

    3x3 MIMO uses three antennas to establish up to three streams of data with the receiving device. Compared to ordinary single antenna (SISO) networks, 3x3 offers up to a 300% increase in throughput.

    With three spatial streams established the data payload is divided across all three antennas and transmitted over the same frequency band. The use of three unique polarisations is rare. Most 3x3 radio systems use either VVV-Pol or VVH-Pol configurations. Antennas are however usually spatially distributed within the device to ensure adequate ECC and isolations.

    3x3 MIMO is not commonly found in cellular mobile or microwave-based systems, instead typically only found in 802.11n and first generation 802.11ac Wave 1 WiFi access points. 3x3 became largely redundant with the quick transition from 2x2 to 4x4 MIMO.

    4x4 MIMO, sometimes referred to as 4T4R, uses four antennas to establish up to four streams of data with the receiving device. Compared to ordinary single antenna (SISO) networks, 4x4 offers up to a 400% increase in throughput.

    With four spatial streams established the data payload is divided across all four antennas and transmitted over the same frequency band. Unlike 2x2 MIMO where it was feasible to simply use two polarisations, the use of four unique polarisations is rare. With reduced isolation and Envelope Correlation Coefficient (ECC), this has significant implications - most notably under clear channel conditions where the construction of physically independent propagation paths are not possible.

    4x4 MIMO equipment is widely available with most smartphones and modems now compatible. Under strong signal conditions 4x4 provides around a 90% improvement over 2x2 MIMO, and under weak conditions as high as 160% faster than 2x2.

    4x4 MIMO block diagram

    8x8 MIMO, often referred to as 8T8R, uses eight antennas to establish up to eight streams of data with the receiving device. Compared to ordinary single antenna (SISO) networks 8x8 offers up to a theoretical 800% increase in throughput, although this is significantly lower in practice.

    With eight spatial streams established the data payload is divided across all eight antennas and transmitted over the same frequency band. Unlike traditional 2x2 MIMO antenna element configurations, 8x8 MIMO does not use eight independent polarisations, instead combining dual polarised elements with spatial separation. Design of 8x8 MIMO antennas requires intensive computer simulation to ensure sufficiently low envelope correlations and inter-port isolation.

    8x8 MIMO block diagram

    Massive MIMO includes a number of multiple antenna configurations including 32T32R, 64T64R, and 128T128R. These systems use a large array of antennas to establish multiple streams of data with multiple devices simultaneously. Massive MIMO uses beamforming technology to construct user dedicated beams which provide increased signal power and reduced interference to multiple users at the same time.

    In an nTmR array, there are n transmitting antenna elements and m receiving elements, with each UE leveraging the entire array or a subset therein. Compared to a legacy 2x2 MIMO LTE network, Massive MIMO typically provides a 500 to 800% increase in total cell throughput under the same radio conditions. Massive MIMO is a key enabler of 5G NR, implemented in all sub-6 GHz 5G networks. The technology is rapidly being introduced across existing 4G networks to provide multi-gigabit throughput without requiring new spectrum.

    Massive MIMO block diagram