Multi-Band Multi-Orbit Satellite Architectures

• Multi-band, multi-orbit architectures integrate satellites across different orbits and frequency bands to enhance throughput and resilience under dynamic weather conditions.
• The system employs distributed matching, beam hopping, and ISAC-based sensing to achieve proportional fairness and efficient resource allocation among ground users.
• Empirical results show nearly ideal throughput, 73% higher per-user performance over S-band-only designs, and robust link availability despite rain attenuation.

A multi-band, multi-orbit system architecture in non-geostationary satellite networks (NTNs) integrates satellites operating at multiple frequency bands across distinct orbital layers. This architectural paradigm addresses scalability and resilience challenges in next-generation NTNs, particularly under dynamic atmospheric conditions such as rainfall attenuation. Leveraging distributed matching, beam hopping, and integrated sensing and communications (ISAC), the architecture enhances downlink throughput, link availability, and resource allocation efficiency while achieving proportional fairness among a massive number of ground users (Leyva-Mayorga et al., 2 Dec 2025).

1. System Composition: Orbital Shells, Cells, and Beam Footprints

The architecture comprises $S$ non-geostationary satellites partitioned into $I$ orbital shells, $\mathcal{S} = \bigcup_{i=0}^{I-1}\mathcal{S}_i$, where each shell $i$ is characterized by:

• Altitude $h_i$ (e.g., LEO at 570 km, VLEO at 200 km)
• Inclination $\delta_i$
• Carrier frequency $f_i$, bandwidth $B_i$, and beam count $\beta_i$
• Satellite count per shell, $S_i=|\mathcal{S}_i|$

The ground region is divided into $I$0 quasi–Earth–fixed cells $I$1, each covering a small and approximately static geographic tile. Each satellite $I$2 projects a moving beam footprint $I$3 at frame $I$4, with $I$5. The maximum slant range from $I$6 to any location in cell $I$7 is denoted as $I$8.

Inter-layer coordination is realized via a distributed many-to-one matching protocol matching each cell to at most one satellite per frame, irrespective of orbital shell. Layer handover is implicitly triggered by matching updates, with no explicit handover signaling required outside of the matching exchange.

2. Frequency Bands and Propagation Physics

Each orbital shell operates at a designated frequency band. In the representative instantiation:

• Shell 0 (LEO) utilizes S-band: $I$9 GHz, $\mathcal{S} = \bigcup_{i=0}^{I-1}\mathcal{S}_i$0 MHz
• Shell 1 (VLEO) utilizes K-band: $\mathcal{S} = \bigcup_{i=0}^{I-1}\mathcal{S}_i$1 GHz, $\mathcal{S} = \bigcup_{i=0}^{I-1}\mathcal{S}_i$2 MHz

The free-space path loss between satellite $\mathcal{S} = \bigcup_{i=0}^{I-1}\mathcal{S}_i$3 and cell $\mathcal{S} = \bigcup_{i=0}^{I-1}\mathcal{S}_i$4 at frame $\mathcal{S} = \bigcup_{i=0}^{I-1}\mathcal{S}_i$5 is given as

$\mathcal{S} = \bigcup_{i=0}^{I-1}\mathcal{S}_i$6

with $\mathcal{S} = \bigcup_{i=0}^{I-1}\mathcal{S}_i$7 the speed of light. The receive link SNR is computed by

$\mathcal{S} = \bigcup_{i=0}^{I-1}\mathcal{S}_i$8

Where $\mathcal{S} = \bigcup_{i=0}^{I-1}\mathcal{S}_i$9 is transmit power, $i$0, $i$1 the antenna gains, $i$2 the pointing loss, $i$3 receiver noise, and $i$4 atmospheric attenuation.

Rainfall is modeled as the principal propagation impairment, with attenuation

$i$5

where $i$6 is rain intensity [mm/h], $i$7 the slant path through the rain layer, and $i$8, $i$9 are frequency and polarization-dependent coefficients. S-band links provide improved rain resilience, whereas K-band delivers significantly greater bandwidth.

3. Beam Hopping, Matching, and Scheduling

The synchronization and assignment of resources are realized through a structured frame system, where each system frame of duration $h_i$0 contains $h_i$1 OFDMA downlink frames, $h_i$2 sensing frames, and $h_i$3 feedback frames.

Beam hopping is orchestrated implicitly via:

• A distributed many-to-one deferred acceptance matching $h_i$4, mapping each satellite's quota $h_i$5 of serving cells to at most $h_i$6 cells and ensuring each cell is matched uniquely: $h_i$7

$h_i$8

$h_i$9

• Each cell and satellite generates a preference list prioritized by estimated per-user rate $\delta_i$0.
• The matching outcome ensures Pareto-optimal and stable assignments under the constraint of no inter-cell interference, with the beam hopping schedule embedded into subsequent resource allocation (Leyva-Mayorga et al., 2 Dec 2025).

4. Distributed Resource Allocation and Local Optimization

Once cell-satellite assignments $\delta_i$1 are updated, each satellite $\delta_i$2 distributes its $\delta_i$3 OFDMA frames (and $\delta_i$4 beams) across the assigned cells $\delta_i$5, allocating $\delta_i$6 frames per cell.

Resource allocation per satellite solves the local convex program: $\delta_i$7 Subject to: $\delta_i$8 This proportional fairness objective is efficiently solved via continuous relaxation and interior-point methods, then rounded to integer solutions. Computational complexity per satellite is $\delta_i$9.

5. Integrated Sensing and Communications (ISAC)

A subset $f_i$0 (those operating in K-band or above) transmits orthogonal pilot symbols ($f_i$1 length) for downlink sensing, each hopped in beam across $f_i$2 beams. The instantaneous SNR for cell $f_i$3 from satellite $f_i$4 in AWGN is estimated via the unbiased maximum-likelihood estimator: $f_i$5 with variance bound $f_i$6.

Rain attenuation is then estimated by comparing the no-rain SNR reference with the observed SNR: $f_i$7

These ISAC-based estimates populate the matching and resource allocation framework, ensuring dynamic adaptation to time-varying atmospheric attenuation. The integration of ISAC enables satellites to update their cell associations and resource distributions every 10 ms frame to track rapidly changing conditions.

6. Performance Metrics and Empirical Results

Key metrics include:

• Instantaneous per-user rate (for $f_i$8 users per cell): $f_i$9
• Average per-user throughput per frame:

$B_i$0

• Outage probability, derived from the CDF of $B_i$1 below threshold

Empirical evaluation considers a two-layer S-/K-band constellation (LEO at 570 km/2 GHz, VLEO at 200 km/20 GHz) over 3960 ground cells in Europe, with rain following a discrete-time Markov process and pilots of $B_i$2 symbols.

The ISAC-powered multi-band, multi-orbit architecture demonstrated:

• Achieved approximately 99% of ideal full-CSI throughput
• 73% higher per-user throughput versus S-band-only designs
• Negligible signaling overhead (one UL OFDMA frame per match cycle)

Combining rain-resistant S-band with high-capacity K-band and leveraging ISAC-powered sensing and distributed matching, the system maintains robust availability and throughput under variable weather. Dynamic adaptation through distributed algorithms provides resilience and operational efficiency (Leyva-Mayorga et al., 2 Dec 2025).

7. Implications and Architectural Significance

The integration of multi-band (S-/K-band) transmission with multi-orbit (LEO/VLEO) deployment, beam hopping, and ISAC-enabled resource management represents a comprehensive response to NTNs' scale and variability. A plausible implication is that this architecture can flexibly support diverse service requirements and geographical environments without specialized handover protocols or centralized orchestration, provided rapid local estimation and matching are sustained. This design achieves Pareto-stable, proportionally fair throughput allocation, and high link reliability even in adverse propagation environments. The realization of nearly ideal throughput suggests a pathway for efficient terrestrial-satellite coexistence and scalable 5G NTN operations (Leyva-Mayorga et al., 2 Dec 2025).