Definition
MIMO (Multiple-Input Multiple-Output) uses multiple antennas to transmit and receive simultaneously. It enhances capacity through spatial multiplexing and improves reliability through diversity gain. 4G LTE commonly uses 2×2 and 4×4 MIMO; 5G NR introduces massive MIMO, with antenna numbers reaching 64/128/256. Combined with beamforming, it forms narrow beams in both horizontal and vertical dimensions, significantly improving spectral efficiency and coverage. MIMO works together with OFDM and modulation to form the core of the 4G/5G air interface.
Types and Gains
- SU-MIMO: Single-user MIMO, where the base station's multiple antennas serve a single terminal. Spatial multiplexing increases single-user throughput; requires the terminal to support multiple antennas.
- MU-MIMO: Multi-user MIMO, where the base station's multiple antennas simultaneously serve multiple terminals, distinguishing users through precoding and increasing cell capacity.
- Massive MIMO: Large-scale antenna arrays (64T64R or more), combined with beamforming to achieve three-dimensional beams and user-level beams, enhancing coverage and throughput; core of 5G Sub-6G and millimeter wave.
Channel Feedback and Precoding
MIMO requires the terminal to report CSI (Channel State Information), RI/PMI/CQI, etc., and the base station performs precoding and scheduling based on the feedback. Open-loop MIMO (such as transmit diversity) does not rely on feedback and is suitable for high-speed mobility; closed-loop MIMO relies on feedback and is suitable for low- to medium-speed scenarios. Massive MIMO obtains channels through channel reciprocity (TDD) or SRS feedback to achieve beamforming and precoding.
Typical Use Cases
MIMO is widely used in 4G/5G cellular, Wi-Fi 6, fixed wireless access, etc. IoT devices can use 2×2 MIMO for higher rates at nearby points; at the edge, diversity gain can improve reliability. Massive MIMO plays a key role in high-capacity scenarios such as dense deployments and stadiums.
SimtoLink's IoT modules support multi-antenna configurations (e.g., 2×2 MIMO), which can increase uplink and downlink rates in areas with good signals; in weak coverage areas, diversity gain can improve connection stability. When selecting modules for customers, SimtoLink considers the application scenario (rate requirements, coverage conditions) to recommend suitable MIMO configurations and assist customers in optimizing antenna layouts and installation locations to fully leverage MIMO performance.
Spatial Multiplexing Needs Enough Rank from the Channel
MIMO is not “add another antenna and get twice the rate.” Spatial multiplexing needs more than one distinguishable path in the channel matrix at the same moment. Only when the rank indicator the device reports is greater than 1 can the base station send different data on different layers. In open ground with a strong direct path and weak reflections, rank is often 1. Extra antennas can only form a beam or provide diversity. Throughput does not double with the antenna count. Diversity sends the same data over several paths and trades rate for reliability. It suits the edge and high speed, which is the opposite goal from multiplexing.
Closed-loop precoding depends on the device’s PMI and CQI. If the feedback is stale, the device implements only open-loop transmit diversity, or reporting is switched off, the base station cannot aim the energy. A TDD massive array can use uplink-downlink reciprocity, but once calibration fails that assumption no longer holds and the beam points the wrong way. All of this happens at the physical layer. It has nothing to do with whether the application retransmits.
Antenna Spacing and Device Capability Have to Hold Together
To receive several layers well, antenna spacing on the device usually needs to reach about half a wavelength, and the polarizations should be one horizontal and one vertical, or plus and minus 45 degrees, so the two branches are not highly correlated. At 2.6 GHz, half a wavelength is about 6 cm. Two antennas pressed against a narrow metal edge stay highly correlated, and RI stays at 1 for a long time. A massive array on the base-station side also needs calibration, power, and baseband capacity for carrier aggregation to keep up. Swap only the antenna panel and leave the feature switch off, and the air interface is still a single layer.
Device category is part of the condition too. A module with only one antenna can fall on only one layer even if the cell is configured for four. If an IoT device leaves one of two antenna branches open, the base station may still schedule two branches. The block error rate rises, the outer loop lowers the modulation order, and the result looks like “many signal bars, almost no rate.” Far from a macro base station, confirm that diversity is actually working before chasing layer count.
Ports Crossed, or Both Branches on the Same Polarization
The easiest field error is crossed antenna ports: the main branch wired to the diversity port, or both antennas on the same polarization. VSWR can still look normal, and the rate still will not rise, because the two branches see almost the same channel. Another error is to carry Wi-Fi MIMO habits over to cellular and ignore the cellular power amplifier’s linear region and absorption limits. Pushing more transmit chains without that headroom makes the uplink worse, because the amplifier has to back off.
A test done only in a shielded box sees rich multipath, so the rank indicator reads high. Out of the box, on a rooftop with a direct path, the layer count drops to 1, and that is easy to call a device fault. At the real installation point, record RI, CQI, and the block errors on each layer, and check whether the two receive levels are close. One level clearly low is usually the feeder, the connector, or the antenna itself, not the scheduler.
When the two receive levels differ by many decibels for a long time, swap the feeders before judging. If the difference follows the port, the antenna or the feeder is at fault. If it stays on the same receiver port, that front end of the module is at fault. Changing only the base-station configuration will not repair device hardware. Spatial-multiplexing layers also depend on how large the payload is. When the packet itself is only a few dozen bytes, the airtime saved by an extra layer is negligible, while the cost of correlation and calibration is still there. For that kind of device, confirm that diversity really lowers the block error rate. Do not chase the rank-indicator number. A shielded box is rich in multipath, so rank reads high. A drop in layers after leaving the box for a rooftop with a direct path is normal. It is not evidence that the module is damaged.