The Non-Monotonic Limits of Dense Antenna Arrays

13 concepts9 papers1 proposal

For decades, wireless system design operated under a straightforward heuristic: if you want higher array gain, finer spatial resolution, or greater network capacity, you add more antennas and pack them closer together. This assumption of monotonic improvement underpins traditional Multiple-Input Multiple-Output (MIMO) systems.

However, as wireless engineering pushes toward next-generation paradigms—such as Extremely Large-Scale MIMO (XL-MIMO), Holographic MIMO (HMIMO), and novel waveguide-fed architectures—this foundational intuition breaks down. Recent research reveals that array performance is fundamentally non-monotonic. Beyond a critical threshold, adding more elements or decreasing their physical spacing causes performance to peak and sharply decline. Understanding why this peak occurs requires looking at the physical, geometric, and network-level trade-offs that govern dense antenna systems.

The Electromagnetic Penalty of Over-Packing Elements

The most immediate barrier to infinite antenna scaling is mutual coupling—the electromagnetic interaction that occurs when closely spaced elements absorb and re-radiate each other's fields.

In near-field XL-MIMO systems, Wang et al. demonstrate that a system's Effective Degrees of Freedom (EDoF)—the number of independent data streams it can simultaneously support—is primarily bounded by its physical aperture size rather than the sheer number of antenna elements. While tightening the spacing between elements initially improves spatial resolution, it eventually triggers severe electrical interference. This mutual coupling causes a distinct peak and subsequent decline in total system capacity, proving that packing more elements into a fixed aperture yields diminishing, and eventually negative, returns.

This phenomenon is not confined to flat, two-dimensional surfaces. When expanding into three-dimensional antenna arrays, Ji et al. show that performance is highly sensitive to the physical spacing between array layers. Designers face a strict optimization challenge: they must find the exact inter-layer spacing that maximizes constructive interference while mitigating the detrimental coupling effects that degrade radiation efficiency.

Similarly, in HMIMO communications, Yuan et al. highlight a fundamental trade-off between individual element efficiency and overall spatial resolution. High antenna density is a prerequisite for achieving the spatial resolution required for superdirectivity, yet the resulting mutual coupling deforms radiation patterns and saps energy efficiency.

Shared-Power Dynamics in Waveguide Architectures

The non-monotonic behavior of array gain becomes uniquely visible in emerging hardware paradigms like Pinching-Antenna Systems (PASS). Developed to bypass the path-loss limitations of over-the-air transmission, PASS utilizes a dielectric waveguide as an "antenna rail" where small radiating particles can be flexibly attached or "pinched" to emit signals.

Traditional Array (Independent Feeds):
[Source] -> [Element 1]
[Source] -> [Element 2]   --> Scaling is often linear/monotonic
[Source] -> [Element 3]

Pinching-Antenna System (Shared Feed):
[Source] ===(Waveguide Rail)===> [Element 1] ===> [Element 2] ===> [Element 3]
                                       \                \                \
                                    (Radiates)       (Radiates)       (Radiates)

Intuitively, one might assume that adding more radiating elements along this waveguide rail would continuously boost signal strength. However, Ouyang et al. mathematically demonstrate that array gain does not increase monotonically with the number of antennas or with decreasing spacing. Instead, an optimal configuration exists for both variables. Crucially, the authors show that the optimal number of antennas actually decreases as the minimum inter-antenna spacing increases.

The physical intuition behind this curve is rooted in energy distribution. As Wang et al. explain, because each radiating element draws power from a single, shared waveguide source, they are inherently coupled. Each additional element alters the remaining power available downstream. Over-populating the waveguide rail causes an inefficient power distribution, creating a non-monotonic relationship where adding elements past the optimum point actively diminishes the total radiated signal strength.

The Spatial Resolution vs. Aliasing Trade-off

Beyond the physics of power and coupling, the geometric layout of an array introduces a structural trade-off between spatial resolution and interference management.

When configuring large arrays, designers often choose between sparse layouts (where elements are spread far apart) and compact layouts (where elements are tightly clustered). As Min et al. explore, the optimal configuration depends entirely on user density:

  • Sparse Geometries: By expanding the outer boundary of the array, sparse layouts maximize the physical aperture, providing superior spatial resolution when serving a small number of isolated users.
  • Compact Geometries: When the network faces a high density of concurrent users, sparse layouts break down due to grating lobe aliasing—spurious interference peaks caused by under-sampling the spatial wavefront. Compact layouts are required in these scenarios to suppress these grating lobes and maintain robust communication.

Because a static array configuration cannot simultaneously optimize for high-resolution sensing and high-capacity, multi-user communication, the authors propose dynamically shifting the active array geometry using structured codebooks to match environmental demands.

Translating Geometry to Network Coverage

The non-monotonic relationship between array configuration and performance ultimately scales up to dictate network-level coverage and scheduling policies. The core tension here lies in balancing antenna beamwidth against user and infrastructure density.

In aerial networks, such as cellular-connected unmanned aerial vehicles (UAVs), Sun et al. demonstrate that the optimal beamwidth for directional antennas is inversely proportional to the density of the ground infrastructure. In sparse ground networks, wide beams are mandatory to maintain a stable connection as the UAV moves. However, in dense ground networks, wide beams collect excessive interference from neighboring cells, making narrow beams necessary to isolate the desired signal.

Low Infrastructure Density:  [Wide Beam = Necessary Coverage]
High Infrastructure Density: [Wide Beam = Severe Inter-Cell Interference] -> Must narrow the beam

A parallel optimization problem exists in millimeter-wave (mmWave) small cells, where highly directional beams are used to combat severe path loss. As Niu et al. argue, relying solely on ultra-narrow, high-gain beams creates massive scheduling overhead when multicasting content to a group of users, as the base station must sequentially sweep across positions. Conversely, an overly wide beam suffers from low antenna gain.

To resolve this, Niu et al. utilize multi-level codebooks to dynamically adjust the antenna beamwidth based on localized user density. By matching the beam's spatial coverage to the physical clustering of users, the system optimizes network throughput.

Conclusion

The shared insight across these diverse research tracks is that advanced antenna arrays cannot be treated as linear systems where "more is always better." Whether limited by mutual coupling in holographic surfaces, power depletion in waveguide rails, grating lobes in sparse XL-MIMO layouts, or interference in dense cellular networks, performance consistently follows a non-monotonic curve. The future of wireless design lies not in building infinitely larger or denser arrays, but in dynamically adapting array configurations, spacings, and beamwidths to hit the exact peak of that curve.

History & citationVersion 1 · 2026-10-04

Current · Permanent link

View frozen version 1

Cite this direction

Latest version (a living page):

@misc{lacuna_non_monotonic_array_gain_5651cb5b,
  title = {non-monotonic array gain},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/non-monotonic-array-gain/rdr_5651cb5bd6064eb896041c5702fa7ddf},
  note = {Latest version of a living research direction; cite a version for a fixed text. Accessed 2026-11-19},
}

This exact version (version 1, frozen text):

@misc{lacuna_non_monotonic_array_gain_5651cb5b_v1,
  title = {non-monotonic array gain},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/non-monotonic-array-gain/rdr_5651cb5bd6064eb896041c5702fa7ddf/v/1},
  note = {Version 1, frozen 2026-10-04. Accessed 2026-11-19},
}
About

These directions overlap with this one through shared papers.

See something wrong? Report a mistake