Statistical CSI Based Hybrid mmWave MIMO-NOMA with Max-Min Fairness

Jinle Zhu, Qiang Li, Hongyang Chen, H. Vincent Poor

TL;DR

This paper introduces a robust framework for mmWave MIMO-NOMA systems that prioritizes "max-min fairness"—ensuring the users with the worst connections still get reliable service. By leveraging long-term Statistical Channel State Information (SCSI) instead of hard-to-track instantaneous data, the authors design a hybrid beamforming and power allocation strategy that balances high spectral efficiency with absolute user fairness.

What Problem Does This Paper Address?

As we move toward 5G-Advanced and 6G, two technologies are front and center: Millimeter Wave (mmWave), which offers massive bandwidth, and Non-Orthogonal Multiple Access (NOMA), which allows multiple users to share the same frequency/time resource.

However, combining them creates two major headaches:

  1. CSI Overhead: mmWave channels change incredibly fast. Trying to get perfect, real-time "Instantaneous CSI" (ICSI) for every user creates a massive communication overhead that can eat up the very bandwidth you're trying to save.
  2. The Fairness Gap: Most existing beamforming designs aim to maximize "Sum Rate" (total system speed). In a NOMA system, this usually results in the strongest users getting all the resources while "weak" users (those far away or behind obstacles) get nothing.

The authors tackle these by asking: How can we design a system that is fair to everyone using only stable, long-term channel statistics?

Methods

The paper proposes a three-stage solution based on SCSI (like the average angles of signal arrival/departure) rather than fast-fading data.

  • User Grouping (Modified AGNES): They use an Agglomerative Nesting (AGNES) clustering algorithm. Instead of physical distance, it groups users based on Correlation Matrix Distance (CMD)—essentially grouping users whose channels "point" in similar directions.
  • Hybrid Beamforming:
    • Analog Stage: Unlike traditional methods that point a sharp beam at the strongest user, their "Statistical Analog Beamforming" (SAB) uses the average covariance of the whole group to create a beam that covers all users in a cluster.
  • Digital Stage: They propose two designs to kill inter-cluster interference: a standard Zero Forcing (ZF) approach and an extended Signal-to-Leakage-plus-Noise Ratio (SLNR) metric. The SLNR is particularly clever as it balances maximizing a group's signal while minimizing the "leakage" interference it causes to other groups.
  • Power Allocation: To solve the "Max-Min" problem (maximizing the minimum rate), they use a Quadratic Transform (QT) to turn a non-convex, "unsolvable" math problem into a convex one, then find the optimal power levels using a bisection search.

Datasets

The study relies on Monte Carlo simulations rather than a static dataset.

  • Setup: A Base Station with 64 antennas and 4 RF chains.
  • Channel Model: A general ray-based directional model with 6 multipath components per user.
  • Environment: Users are randomly distributed with Angles of Departure (AoD) uniformly spread across a 180-degree sector.

Key Insights and Ideas

  • Fairness is a Beamforming Problem: You can't fix fairness with power allocation alone if your beam is physically pointing away from the weak user. By designing the analog beam to represent the average of the cluster, the authors provide a "baseline" gain that power allocation can then refine.
  • The Power of SCSI: The paper proves that long-term statistics (which are much easier to measure) are "good enough" for high-performance NOMA. This significantly reduces the feedback overhead compared to ICSI-based systems.
  • SLNR for Clusters: While SLNR is common for single users, extending it to NOMA clusters allows the system to handle "overloaded" scenarios (more users than RF chains) more gracefully than simple Zero Forcing.

Key Results

  • Superior Fairness: In simulations (K=9 users, SNR=0 dB), the proposed SCSI-based algorithms significantly outperformed "cluster-head" designs, which tend to ignore weak users.
  • Stability: The minimum data rate achieved by their method remains more stable as the number of users increases compared to K-means clustering or ICSI-based methods.
  • Efficiency: The SLNR-based digital beamforming achieved nearly the same performance as Zero Forcing but proved more effective at suppressing interference in complex, multi-user environments.

Limitations and Future Work

  • Computational Complexity: While the SCSI approach reduces communication overhead, the iterative power allocation (QT + Bisection) still requires significant processing power at the base station.
  • Stationary Assumption: The model assumes "Wide-Dense Stationary" statistics. In high-mobility scenarios (like a fast-moving vehicle), even these "long-term" statistics might change too rapidly, requiring further adaptation.
  • Perfect Hardware: The study assumes ideal hardware; real-world phase shifters in hybrid beamforming often have "quantization errors" that could degrade the precision of these statistical beams.

Summary updated

Abstract

Non-orthogonal multiple access (NOMA) and millimeter wave (mmWave) are two key enabling technologies for the fifth-generation (5G) mobile networks and beyond. In this paper, we consider mmWave NOMA systems with max-min fairness constraints. On the one hand, existing beamforming designs aiming at maximizing the spectrum efficiency (SE) are unsuitable for the NOMA systems with fairness in this paper. On the other hand, previous work on about mmWave NOMA mostly depends on full knowledge of channel state information (CSI) which is extremely difficult to obtain accurately in mmWave communication systems. To address this problem, we propose a heuristic hybrid beamforming design based on the statistical CSI (SCSI) user grouping strategy. An analog beamforming scheme is first proposed to integrate the whole cluster users to mitigate the inter-cluster interference in the first stage. Then two digital beamforming designs are proposed to further suppress the interference based on SCSI. One is the widely used zero forcing (ZF) approach and the other is derived from the signal-to leakage-plus-noise ratio (SLNR) metric extended from orthogonal multiple access (OMA) systems. The effective gains fed back from the users are used for the power allocation. We introduce the quadratic transform (QT) method and bisection approach to reformulate this complex problem so as to rend it solvable. Simulation results show that our proposed algorithms outperform the previous algorithms in term of user fairness.
BibTeX
@inproceedings{zhu2021statistical,
  author = {Jinle Zhu and Qiang Li and Hongyang Chen and H. Vincent Poor},
  title = {{Statistical CSI Based Hybrid mmWave MIMO-NOMA with Max-Min Fairness}},
  booktitle = {ICC},
  year = {2021},
  doi = {10.1109/icc42927.2021.9500676},
}

Loading related content...

Figure