Multiuser MIMO (MU-MIMO)

The core physical-layer mechanism enabling MU-MIMO. Instead of dividing resources by time or frequency, the access point uses precoding to create spatially distinct beams. Each beam is directed at a specific user, ensuring that the signal intended for User A appears as a null or negligible interference at User B. This relies on the access point having accurate Channel State Information (CSI) for all connected terminals.

The primary engineering challenge in MU-MIMO is managing interference between users without coordination between their receivers. This is achieved through transmit-side precoding algorithms:

• Zero-Forcing (ZF): Completely nulls interference but can amplify noise.
• Block Diagonalization (BD): Forces the aggregate channel matrix into a block-diagonal form, decoupling users.
• Dirty Paper Coding (DPC): A theoretical optimal non-linear technique that pre-subtracts known interference.

MU-MIMO operates differently depending on the link direction:

• Downlink (Broadcast Channel): The access point transmits to multiple users. Requires complex precoding at the transmitter to separate data streams.
• Uplink (Multiple Access Channel): Multiple users transmit to the access point. The base station uses joint detection algorithms like Successive Interference Cancellation (SIC) to separate the overlapping signals, which is less computationally intensive than downlink precoding.

MU-MIMO performance is critically dependent on the freshness and accuracy of Channel State Information (CSI). In systems like 802.11ac/ax, the access point initiates an explicit sounding sequence:

1. The AP transmits a Null Data Packet (NDP).
2. Each user estimates its channel and sends back a compressed Beamforming Report. This overhead scales with the number of users and antennas, creating a trade-off between spatial multiplexing gains and protocol efficiency.

Not all user groups can be served simultaneously. The scheduler must select a subset of users whose channels are sufficiently uncorrelated. Key considerations include:

• Spatial Correlation: Users with highly correlated channel vectors (e.g., co-located devices) cannot be separated spatially and must be time-multiplexed.
• Channel Condition Number: A well-conditioned aggregate channel matrix is required for stable precoding.
• Fairness vs. Throughput: Maximizing sum-rate often favors users with strong channels, requiring weighted scheduling algorithms to ensure fairness.

MU-MIMO scales to its logical extreme in Massive MIMO, where base stations employ hundreds of antennas to serve tens of users. In this regime, the law of large numbers causes random channel vectors to become nearly orthogonal, a phenomenon called favorable propagation. This simplifies precoding to basic Maximum Ratio Transmission (MRT) and dramatically increases spectral efficiency, making it a foundational technology for 5G NR.

A linear precoding technique for MU-MIMO that eliminates inter-user interference by constraining each user's precoding matrix to lie in the null space of all other users' channel matrices.

• Zero Interference: Completely orthogonalizes spatial streams between users
• Dimensionality Requirement: Requires the transmitter to have more antennas than the sum of all receivers' antennas
• Practical Trade-off: Simpler than Dirty Paper Coding but suboptimal in low-SNR regimes

A non-linear detection technique that iteratively decodes the strongest signal stream, subtracts its reconstructed contribution from the received signal, and repeats the process for remaining streams.

• Ordering Matters: Decoding order significantly impacts performance; strongest signal decoded first
• Error Propagation: Incorrect decoding of an early stream corrupts all subsequent cancellations
• NOMA Integration: Forms the basis of Power-Domain NOMA in 5G, where users are separated by power levels