TL;DR
This paper proposes using Non-Orthogonal Multiple Access (NOMA) to allow far-field users to "hitch a ride" on the spatial beams already focused on near-field users in massive MIMO networks. By sharing these preconfigured beams, the system can serve additional users and increase total data rates without requiring new, dedicated spectral resources.
What Problem Does This Paper Address?
As we move toward 6G, wireless networks are using higher frequencies (like mmWave and THz) and massive antenna arrays. This combination significantly extends the Rayleigh distance—the boundary where the "near-field" ends and the "far-field" begins.
In the near-field, radio waves are spherical, requiring beam-focusing (targeting a specific 3D coordinate). In the far-field, waves are planar, requiring beam-steering (targeting a 2D direction). Traditionally, these two types of communications are treated separately. The authors address the challenge of how a "legacy" network, already optimized for near-field users, can efficiently admit additional far-field users without disrupting existing services or wasting bandwidth.
Methods
The authors propose a NOMA-based framework where a base station (BS) uses its existing near-field beams to simultaneously serve far-field users.
- NOMA Integration: Because near-field users are typically closer to the BS and have stronger channel conditions, they are tasked with performing Successive Interference Cancellation (SIC). They decode and subtract the far-field user's signal before decoding their own. Far-field users, having weaker signals, simply decode their data while treating the near-field signal as noise.
- Resource Allocation: The authors formulate an optimization problem to maximize the sum data rate of the far-field users while guaranteeing a minimum Quality-of-Service (QoS) for the legacy near-field users.
- SCA Algorithm: To solve the complex, non-convex optimization problem, they propose a Successive Convex Approximation (SCA) algorithm. This iteratively linearizes the problem into a convex form that can be solved efficiently.
- Branch-and-Bound (BB): For specific simplified cases (e.g., a single far-field user), they use a Branch-and-Bound algorithm to find the globally optimal performance ceiling for comparison.
Datasets
The paper relies on numerical simulations rather than physical datasets. The channel models are based on established electromagnetic theory:
- Near-Field: Modeled using the spherical-wave propagation model.
- Far-Field: Modeled using the conventional planar-wave beamsteering vector.
- Parameters: Simulations used a carrier frequency of 28 GHz, a 36-user setup, and Uniform Linear Arrays (ULA) with 64 or 128 antennas.
Key Insights and Ideas
- Beam Reuse: The core "aha moment" is that even though near-field beams are tightly focused on a specific point, they still possess enough spatial characteristics to be useful for far-field users located in the same angular direction.
- Antenna Scaling: The performance of this NOMA coexistence improves significantly as the number of antennas at the base station increases. More antennas allow for sharper near-field focusing, which paradoxically creates better opportunities for NOMA-based sharing.
- Robustness: Even with imperfect Channel State Information (CSI)—where the base station doesn't have a perfect "map" of the far-field users—the proposed NOMA scheme still outperforms traditional greedy scheduling baselines.
Key Results
- Sum Rate Gains: The NOMA approach consistently achieves higher sum data rates for far-field users compared to baseline greedy allocation schemes.
- Antenna Impact: Moving from 64 to 128 antennas at the base station resulted in a visible upward shift in the sum rate curves (e.g., at 30 dBm transmit power, the sum rate increased by approximately 1-2 bits per channel use).
- Optimal User Count: The simulations revealed that for a fixed number of available beams, there is an "optimal" number of far-field users ($K$) that maximizes the total system throughput.
Limitations and Future Work
- Single-Antenna Users: The study assumes each user has only one antenna (SISO). Real-world 6G devices will likely have multiple antennas, suggesting a need for MIMO-NOMA research in this context.
- Perfect SIC: The model assumes near-field users can perform SIC perfectly. In practice, residual interference from imperfect SIC can degrade performance.
- Deterministic vs. Random: While the paper provides a deterministic study for optimality, real-world deployments involve highly dynamic and random user movements that may require more robust scheduling algorithms.
Summary updated