Omnicell Wireless Communication System

• Omnicell Wireless Communication System is a cellular architecture that uses a full-angle Ray Antenna Array to achieve 360° uniform coverage and cost-efficient design.
• It replaces conventional sectoring with multiple sULAs oriented to deliver narrow, non-overlapping beams for reduced inter-user interference.
• The system offers practical benefits such as improved multiuser sum-rate performance, scalable deployment, and lower hardware costs in mmWave and THz applications.

The Omnicell Wireless Communication System is a cellular architecture in which a base station is deployed at the geometric center of a cell and equipped with a full-angle Ray Antenna Array (RAA) that provides uniform angular resolution over the full azimuth domain. In the formulation reported in "Full-Angle Ray Antenna Array and Omnicell Wireless Communication System" (Zhu et al., 6 Sep 2025), the system departs from conventional cell sectoring based on ULA/UCA hybrid analog/digital beamforming by replacing sectorized sweeping and per-element phase shifting with many oriented simple uniform linear arrays (sULAs), ray selection, and baseband combining. The resulting paradigm is designed to provide uniform $360^\circ$ coverage, high beamforming gain, reduced inter-user interference, and substantially lower hardware cost.

1. Conceptual basis and cellular paradigm

The omnicell paradigm replaces the conventional sectorized deployment, such as three $120^\circ$ sectors served by ULA/UCA arrays, with a single base station located at the center of each cell. That base station uses a full-angle RAA whose rays span the full azimuth range $[-\pi,\pi]$. Users in any direction are associated with one of many rays oriented toward their azimuth angle, so the architecture avoids sector boundaries and the overlapping beams that degrade edge-user performance in conventional sectorized systems (Zhu et al., 6 Sep 2025).

In this design, the central structural distinction from ULA/UCA sectoring is that each ray is a simple uniform linear array whose beam is determined by physical orientation rather than by intra-array phase-shifter control. Conventional ULA/UCA sectoring relies on phase shifters and codebooks to sweep sectors, and the ULA beampattern samples $\sin(\phi)$ uniformly, which yields non-uniform angular resolution at large $|\phi|$ because of the nonlinearity of $\sin(\cdot)$. By contrast, the full-angle RAA builds many sULAs with different orientations $\eta_n$ spanning the full azimuth and uses selection or switching across rays together with simple baseband combining.

The paper identifies two principal expected benefits. First, inter-user interference is reduced because the main lobes associated with different rays are narrow and non-overlapping, while directional elements suppress leakage outside the intended angular region. Second, cost efficiency improves because the RAA uses no per-element phase shifters. The reported use cases include UAV swarms and environment sensing in integrated sensing and communications (ISAC), where uniform high spatial resolution and large sum-rate are required under low transmit SNR and strict cost constraints.

2. Full-angle RAA architecture and geometry

The full-angle RAA consists of $N$ rays, each ray being an sULA with $M$ antenna elements spaced by $\lambda/2$, where $120^\circ$0 is the carrier wavelength. The first element of each ray is displaced by a radial offset $120^\circ$1 from the origin so that adjacent first elements maintain at least $120^\circ$2 separation. The orientation angles $120^\circ$3 span the full angle domain $120^\circ$4, with index set

$120^\circ$5

For the $120^\circ$6th element of the $120^\circ$7th ray, the position is

$120^\circ$8

with

$120^\circ$9

This geometry produces identical intra-ray spacing across all rays while allowing the array to sample the full azimuth by changing only the ray orientations (Zhu et al., 6 Sep 2025).

Under a uniform plane wave with angle of arrival $[-\pi,\pi]$0 and direction vector $[-\pi,\pi]$1, the relative angle to the $[-\pi,\pi]$2th ray is

$[-\pi,\pi]$3

The sULA array response vector for the $[-\pi,\pi]$4th ray is

$[-\pi,\pi]$5

The first-element scalar term is

$[-\pi,\pi]$6

where $[-\pi,\pi]$7 is the element radiation pattern.

After directly summing the $[-\pi,\pi]$8 elements in each ray, the per-ray output vector is

$[-\pi,\pi]$9

The paper characterizes this as a Dirichlet kernel in $\sin(\phi)$0, constant across rays because of the uniform intra-sULA geometry.

A central design rule is the orientation set

$\sin(\phi)$1

which aligns the null of one ray’s main lobe with the peak of its neighbors. The minimum radial offset required for $\sin(\phi)$2 separation among adjacent first elements is

$\sin(\phi)$3

The corresponding ray-count expression for full-angle coverage is

$\sin(\phi)$4

For $\sin(\phi)$5, the paper gives the approximations

$\sin(\phi)$6

3. Beamforming model and spatial resolution

A defining feature of the architecture is that there is no intra-sULA beamforming. All $\sin(\phi)$7 elements in a ray are hard-connected and summed, so each ray produces a natural beampattern without requiring phase shifters across the $\sin(\phi)$8 elements. Beam management is moved to the inter-ray level: a ray selection network (RSN) connects $\sin(\phi)$9 RF chains to $|\phi|$0 rays through a binary selection matrix

$|\phi|$1

with constraints

$|\phi|$2

Selected rays are then combined for user $|\phi|$3 through a baseband beamforming vector $|\phi|$4 (Zhu et al., 6 Sep 2025).

The paper states that the full-angle RAA beamwidth is constant across all directions:

$|\phi|$5

This is the core of the claimed uniform angular resolution. Because the relative angle $|\phi|$6 is sampled uniformly through the orientation increment $|\phi|$7 and because every ray has the same Dirichlet-kernel form in $|\phi|$8, the main-lobe width does not depend on the global azimuth $|\phi|$9.

The null-peak alignment criterion follows directly from the ray beampattern. The nulls satisfy

$\sin(\cdot)$0

and choosing

$\sin(\cdot)$1

places the main-lobe peak of one ray at the first null of its adjacent ray. This suppresses mutual leakage between neighboring rays and is one of the principal mechanisms by which the omnicell architecture reduces interference.

For comparison, the paper gives the ULA hybrid-beamforming response

$\sin(\cdot)$2

and

$\sin(\cdot)$3

The UCA benchmark is also parameterized in the paper through

$\sin(\cdot)$4

with

$\sin(\cdot)$5

The analytical spatial-resolution metric used in the paper is the main-lobe width $\sin(\cdot)$6 together with the null-peak alignment condition. The paper does not provide CRLB or DoA estimation bounds. A plausible implication is that the treatment is oriented toward communication-array performance rather than parameter-estimation optimality.

4. Channel model, SINR formulation, and achievable rate

The evaluations are carried out under a far-field uniform plane wave assumption. Although the paper references near-field XL-MIMO literature, the RAA model studied in the reported results remains a narrowband far-field model. For user $\sin(\cdot)$7, the channel is

$\sin(\cdot)$8

with

$\sin(\cdot)$9

The achievable sum-rate is

$\eta_n$0

For the omnicell RAA system, the paper gives

$\eta_n$1

The noise is modeled as

$\eta_n$2

so the received noise power is $\eta_n$3, with $\eta_n$4.

For sectorized ULA/UCA systems, the paper gives

$\eta_n$5

where $\eta_n$6 is the DFT codebook for ULA or the parametric codebook for UCA, and the channel is restricted to the sector $\eta_n$7:

$\eta_n$8

$\eta_n$9

Baseband beamforming is evaluated through MMSE equalization:

$N$0

with

$N$1

The selection matrices $N$2 and $N$3 are optimized to maximize $N$4 under the stated selection constraints.

The simulation channel model follows the 3GPP Uma NLOS scenario in TS 38.901, using the parameter set reported in the paper: central frequency $N$5 GHz, number of users $N$6, number of RF chains $N$7, number of clusters $N$8, number of rays per cluster $N$9, and thus total paths per user $M$0. The LOS azimuth obeys $M$1, and the transmit SNR per user spans $M$2 dB to $M$3 dB in the multiuser tests.

The directional element model is also taken from 3GPP:

$M$4

$M$5

with

$M$6

According to the paper, these directional elements allow each sULA to focus energy in its narrow responsibility region and thereby increase effective gain relative to isotropic UCA elements.

5. Comparative performance and hardware cost

The paper reports both beampattern-level and system-level comparisons against conventional sectorized architectures (Zhu et al., 6 Sep 2025). In a representative full-angle RAA beampattern example at $M$7 GHz with $M$8, $M$9, $\lambda/2$0 mm, and $\lambda/2$1, the full-angle RAA exhibits constant beamwidth and higher main-lobe amplitude than ULA hybrid beamforming at large $\lambda/2$2, specifically near $\lambda/2$3 or $\lambda/2$4. The corresponding ULA benchmark uses three ULAs, each with $\lambda/2$5 serving a $\lambda/2$6 sector. The paper attributes the ULA degradation to wider beamwidth at large $\lambda/2$7 and lobe overlap across sector boundaries, which increases inter-sector interference.

The UCA sectoring comparison uses $\lambda/2$8 isotropic elements on a circle of radius $\lambda/2$9 mm. The paper states that UCA can achieve uniform angular resolution, but it still requires per-element beamforming and exhibits lower power gain than the directional-element RAA.

In multiuser simulation under 3GPP Uma NLOS, the omnicell RAA configuration uses $120^\circ$00 and $120^\circ$01, the ULA sectoring benchmark uses a $120^\circ$02-element ULA per sector, and the UCA sectoring benchmark uses a $120^\circ$03-element UCA. With $120^\circ$04, $120^\circ$05, and MMSE combining, the omnicell RAA achieves larger sum-rate than both ULA and UCA sectoring across transmit SNR from $120^\circ$06 dB to $120^\circ$07 dB. The reported reasons are higher directional gain per ray, constant beamwidth, null-peak alignment, and greedy selection-matrix optimization framed as a minimum angle-distance problem.

The paper also provides explicit cost models. For RAA,

$120^\circ$08

For ULA sectoring,

$120^\circ$09

The mmWave component benchmarks cited are a phase shifter (ADMV4728) costing more than $120^\circ$1028.62 per unit, and a cheap antenna element costing at most $0.01 per unit in the paper’s model.

The reported example totals are $120^\circ$1176,801 for ULA sectoring, implying that omnicell RAA reduces hardware cost to $120^\circ$12 of ULA sectoring while achieving better performance. The same section states that omnicell is more cost-efficient when antenna element price is at most $120^\circ$131\%$ of the hardware cost of classical HBF.

6. Deployment guidelines, limitations, and extensions

The deployment guidance in the paper is tightly linked to the array geometry and the central-cell architecture. Base-station placement at the cell center is recommended in order to exploit uniform angular resolution and symmetric coverage while avoiding sector boundaries and overlaps (Zhu et al., 6 Sep 2025). Parameter selection is expressed in terms of beamwidth and spacing constraints: choose $120^\circ$14 according to the target beamwidth

$120^\circ$15

set $120^\circ$16 using

$120^\circ$17

determine $120^\circ$18 from the coverage constraint, and use the orientation set

$120^\circ$19

to ensure neighbor null-peak alignment. The paper further recommends directional elements consistent with the 3GPP pattern and MMSE baseband combining together with RSN optimization.

The architecture is presented as especially suitable for mmWave, exemplified by $120^\circ$20 GHz in simulation, and even THz operation, because the wavelength becomes small and the avoidance of expensive phase shifters is particularly consequential. The paper also states that the 2D full-angle RAA covers azimuth and that an extension to full solid-angle 3D coverage could be built by arranging sULAs on a spherical or conical surface with designed elevation and azimuth orientations. This suggests a natural pathway from azimuth-only omnicell coverage to volumetric beam management.

Several limitations are explicit. The present model assumes narrowband uniform plane waves, so wideband beam squint is not analyzed. The paper identifies future work on true time-delay networks or broadband RAA variants for wideband mmWave/THz settings. Near-field effects are likewise outside the reported evaluation: for very large apertures or close users, spherical wavefronts would require a different model. Mobility and tracking are also open issues, since rapid RSN selection updates and switching losses may become important under fast user movement.

Practical implementation constraints include calibration, mutual coupling, and manufacturing tolerances. The paper states that RSN connectivity and per-ray path length must be calibrated to maintain coherent summation and beampattern fidelity. The spacing constraints on $120^\circ$21 and $120^\circ$22 are intended to mitigate coupling by ensuring at least $120^\circ$23 separation, but dense radial packing may still introduce coupling effects, so comprehensive electromagnetic design and decoupling measures are needed. Orientation placement, radial offset, and $120^\circ$24 spacing must be controlled tightly, and the directional-element patterns should conform to the 3GPP model to realize the expected gains.

Taken together, these points situate the omnicell wireless communication system as a full-angle, ray-oriented alternative to sectorized hybrid beamforming. Its principal technical claims are uniform angular resolution over $120^\circ$25, beamwidth equal to $120^\circ$26, elimination of per-element phase shifters, and improved multiuser sum-rate and hardware cost relative to the ULA/UCA sectoring baselines evaluated in the paper.