Multi-Functional Satellite Systems

• Multi-Functional Satellite Systems are integrated platforms that merge communications, sensing, and PNT functions into unified or coordinated architectures.
• They leverage joint waveform design, distributed MIMO, and cooperative inter-satellite links to optimize spectral efficiency and resource utilization.
• MFSS architectures enhance performance through dynamic beamforming, autonomous control, and multi-objective optimization across mission domains.

Multi-Functional Satellite Systems (MFSS) are satellite systems that integrate two or more of the core space services—communications, sensing, and positioning–navigation–timing (PNT)—into a single, unified payload, or into a tightly coordinated multi-satellite infrastructure that shares hardware, spectrum, processing, and orbital resources. In the 6G and non-terrestrial network literature, MFSS are motivated by the cost, spectral-efficiency, launch-mass, and sustainability limits of maintaining separate constellations for broadband, Earth observation, and PNT, and by the possibility of functional synergy when these services are co-designed rather than merely co-located (Sheemar et al., 30 Sep 2025).

1. Conceptual scope and taxonomy

The contemporary MFSS literature organizes the field along two orthogonal axes. The first axis is the service combination: joint communications and sensing (JCAS), joint communications and PNT (JCAP), joint sensing and PNT (JSAP), and fully integrated joint communications, sensing, and PNT (JCSAP). The second axis is the level of integration: cooperative payloads, integrated payloads, and joint payloads. Cooperative payloads share a platform and some hardware but retain distinct signals and often distinct bands; integrated payloads share band and hardware but separate functions in time, space, or code; joint payloads use a single joint waveform concurrently for multiple functions (Sheemar et al., 30 Sep 2025).

The literature also treats integrated sensing and communications (ISAC) and integrated navigation and communication (INAC) as particularly important specializations. ISAC emphasizes common waveform, hardware, and spectrum for communications and sensing, whereas INAC formalizes the joint provision of navigation/PNT and data communications on a common RF and system platform. In both cases, the key shift is from coexistence to joint engineering of signals, beams, and resource allocation (You et al., 2024, Hou et al., 23 May 2026).

MFSS configuration	Integrated services	Representative mechanism
JCAS	Communications + sensing	Joint waveform and beamforming
JCAP	Communications + PNT	Shared ranging/data signals
JSAP	Sensing + PNT	Shared timing and observation geometry
JCSAP	Communications + sensing + PNT	Unified payload and multi-objective design

This taxonomy is significant because it clarifies that MFSS is not restricted to a single hardware pattern. A system may be multi-functional because one satellite carries a unified payload, because a formation synthesizes a virtual aperture, or because a constellation-level scheduler dynamically assigns spacecraft to communication, sensing, relay, or PNT roles. The architectural commonality is not the spacecraft form factor but the presence of shared physical and logical resources across traditionally separate mission domains.

2. Architectural substrates and shared infrastructure

MFSS architectures rely on a small set of recurring technical substrates: large antenna apertures, inter-satellite connectivity, reconfigurable beamforming, and virtualized payload functions. In massive-MIMO LEO ISAC studies, the reference architecture is a satellite acting as a “flying space base station” with a large multi-antenna payload, often under hybrid analog/digital beamforming because fully digital massive MIMO would require too many RF chains and too much power. The same literature also introduces software-defined payloads, software-defined networking, and reconfigurable intelligent surfaces (RIS) as mechanisms for dynamically reassigning beam patterns, waveforms, and service priorities over shared hardware (You et al., 2024).

At the constellation layer, the multi-satellite MIMO survey formalizes MFSS-enabling cooperation through inter-satellite links (ISLs), which permit either raw-signal exchange, statistical coordination, or higher-level state dissemination among GEO, MEO, LEO, HEO, swarm, and distributed-satellite architectures. In this formulation, multiple satellites create a virtual MIMO channel and can trade spatial multiplexing, diversity, and beamforming gains against synchronization, ISL overhead, and payload complexity. The same survey emphasizes that distributed virtual MIMO is particularly relevant for direct-to-device links, where user terminals cannot provide large antenna gains and the space segment must compensate through coordinated transmission or reception (Bakhsh et al., 2024).

A more stringent form of MFSS infrastructure appears in distributed-aperture studies. In one control-aware beamforming model, a formation of $4$ to $576$ satellites, with a fixed total of $2304$ radiating elements, synthesizes a large virtual antenna whose aperture area increases from approximately $330\ \mathrm{m}^2$ to $730\ \mathrm{m}^2$. The satellites are arranged in circular micro-orbits approximating a Logarithmic Spiral Array in the LVLH frame, with operation studied at 600 km altitude and a D2C beam footprint of radius 3 km at 1 GHz. This architecture demonstrates that MFSS can be realized as a distributed RF aperture rather than as a monolithic payload, but it also shows that such apertures are inseparable from guidance, navigation, and control: beamforming weights depend explicitly on perturbed translations and attitudes, via element positions

$$\mathbf{p}_{n,e}= \mathbf{t}'_{n \rightarrow 0} + \mathbf{R}'_{n \rightarrow 0}\,\mathbf{d}_e,$$

so orbit and attitude estimation become part of the signal-processing chain (Tuzi et al., 13 Oct 2025).

The architectural implication is that MFSS should be understood as a systems problem spanning payload design, ISL topology, control, and network orchestration. Shared spectrum without shared timing, shared apertures without state calibration, or shared waveforms without reconfigurable resource management do not produce the full functional gains claimed for MFSS.

3. Cooperative communications and distributed MIMO functions

The communications literature provides some of the most explicit physical-layer realizations of MFSS. In “Space MIMO,” a single-antenna unmodified handheld uplink is received simultaneously by a cluster of LEO satellites, which cooperate through ISLs and a fusion satellite to form an effective distributed SIMO receiver: $\mathbf{y} = \sqrt{p}\,\mathbf{h}s + \mathbf{n}.$ Two CSI regimes are treated. In the full-CSI case, satellites forward received samples and instantaneous channel estimates to the fusion node, which applies a robust linear MMSE combiner. In the partial-CSI case, satellites locally equalize,

$y'_m = \frac{y_m}{\hat{h}_m},$

and forward only preprocessed streams plus long-term statistics, reducing real-time coordination overhead. For a Starlink-like constellation of $3168$ satellites in two shells, with at least $28$ satellites above $576$0 elevation for a user near London, the study reports a capacity of $576$1 Mbits/sec through cooperation of $576$2 satellites with occupied bandwidth of $576$3 MHz, whereas a conventional nearest-satellite link with the same parameters yields less than $576$4 Mbits/sec; with $576$5 cooperating satellites, the reported capacity exceeds $576$6 Gbit/s (Omid et al., 2023).

The same paper is important for MFSS because it identifies the coordination granularity at which shared infrastructure becomes operationally useful. Full CSI maximizes capacity but requires per-symbol exchange of complex samples and per-coherence exchange of instantaneous CSI. Partial CSI shifts processing toward the edge, reduces dependence on real-time CSI exchange, and preserves most of the gains. In MFSS terms, this is the difference between an architecture that spends its shared resources on cooperation overhead and one that uses local intelligence to preserve scalability.

A downlink counterpart appears in recent multi-satellite massive-MIMO work, where multiple LEO satellites transmit independent data streams to multi-antenna user terminals under statistical CSI. The central optimization problem is weighted sum-rate maximization under per-satellite power constraints, reformulated through multi-satellite WMMSE. To avoid iterative onboard optimization and large inter-satellite signaling overhead, the paper introduces a learning-based WMMSE design with tensor equivariance and closed-form recovery, together with a decentralized scheme in which each satellite infers its local precoders from local statistical CSI and periodically available satellite state information such as orbital positions and attitude. Numerical results show that multi-satellite transmission significantly outperforms single-satellite systems in sum rate, and that the decentralized scheme remains close to centralized coordination while substantially reducing computational complexity and inter-satellite overhead (Cao et al., 21 Mar 2026).

Taken together, these results establish a communications-centric MFSS principle: the constellation is no longer a set of independent relay nodes but a programmable distributed antenna system whose functional mode—uplink combining, non-coherent downlink, or partially decentralized cooperation—can be selected according to ISL budget, CSI availability, and service priority.

4. Integrated sensing and communications

JCAS is the most extensively developed MFSS subclass. In massive-MIMO LEO ISAC surveys, the central design variable is a dual-functional waveform and beamformer that simultaneously serves user links and sensing tasks under high Doppler, large delay, and hardware constraints. The canonical hybrid-beamforming model writes the transmit signal as

$576$7

with a weighted optimization balancing communication energy efficiency against sensing beampattern quality (You et al., 2024).

A more concrete formulation appears in rate-splitting-based dual-functional radar-communication satellite systems. In one multibeam LEO DFRC model, a satellite with $576$8 feeds serves $576$9 single-antenna users while simultaneously illuminating a moving ground/sea target for a separate bistatic radar receiver. Messages are split into a common stream and private streams, and the beamforming matrix is optimized to minimize the Cramér–Rao bound (CRB) subject to user QoS and per-feed power constraints. The results show that RSMA-assisted DFRC beamforming outperforms SDMA in the communication-sensing trade-off and in target estimation performance (Yin et al., 2022).

A bistatic LEO ISAC extension strengthens the MFSS interpretation by separating the radar receiver from the satellite and optimizing dual-functional precoders to maximize the minimum user rate under CRB constraints. The same transmitted signal serves $2304$0 communication users and a radar target, and the paper identifies three specific roles for the common stream: beamforming toward the radar target, interference management between communications and radar, and interference management among communication users. Numerical results report a $2304$1–$2304$2 higher minimum user rate than an SDMA-based ISAC baseline with radar sequence and SIC, while the bistatic geometry reduces echo path loss relative to the monostatic case and makes LEO radar sensing more power-feasible (Park et al., 2024).

MFSS scalability becomes explicit in cooperative multi-satellite ISAC networks, where multiple satellites jointly provide downlink communication to UEs and multistatic sensing of airborne targets through multiple gateways. Two sensing architectures are proposed. In the centralized framework, gateways forward raw sensing observations to a central unit, which estimates multiple target positions through a sparse-recovery formulation. In the distributed framework, each gateway performs local estimation and transmits only local position estimates, after which the CU solves data association via the Hungarian algorithm and fuses positions by minimizing distance-estimation error. Both frameworks outperform existing sensing schemes while satisfying communication requirements, and the paper explicitly evaluates the sensing–communication trade-off in terms of sensing accuracy and communication power consumption (Kim et al., 8 Mar 2026).

A related dual-function LEO constellation framework further shows that multiple satellites can jointly optimize communication beamforming and sensing waveform to minimize target-localization CRB under UE rate constraints, using the same hardware and spectrum. This supports the broader MFSS claim that cooperative sensing is not merely an add-on to communication constellations but can be built directly into constellation-level beam and waveform design (Wang et al., 13 Jan 2025).

5. Navigation, PNT, and remote-sensing convergence

JCAP and JSAP introduce a different MFSS coupling structure because timing, geometry, and estimation accuracy become primary resources. INAC expresses this most directly through signal superposition: $2304$3 where navigation and communication share the same channel, spectrum, and hardware, and the power split $2304$4 controls the trade-off between ergodic rate and positioning performance. RIS-aided INAC models further embed this trade-off in the link capacity,

$2304$5

while positioning quality is discussed through PDoP behavior. The same work argues that LEO signals are approximately 30 dB stronger than MEO signals, reports OneWeb and Starlink visibility and positioning behavior under high elevation masks, and notes that LEO-based corrections can reduce PPP convergence from about 30 minutes to less than 2 minutes with sub-meter accuracy (Hou et al., 23 May 2026).

The navigation–remote-sensing branch of MFSS uses analogous machinery but with different metrics. In a dual-function LEO constellation with a Walker Delta architecture of $2304$6 satellites, navigation and remote sensing share a unified frame with a synchronization subframe and a common data subframe. The joint beamforming design minimizes an average weighted PVT error for multiple navigation UEs while satisfying a minimum signal-to-ambiguity-interference-noise ratio (SAINR) for remote sensing. Navigation performance is quantified through CRB-based positioning, velocity, and timing errors; sensing performance is quantified through SAINR after receive beamforming. The optimization explicitly couples navigation beams and sensing beams because the sensing signal acts as interference at navigation UEs and navigation beams contribute interference to the sensing receiver (Wang et al., 16 Nov 2025).

These studies are consistent with the survey taxonomy: JCAP uses shared communication and PNT signals, JSAP uses shared timing and observation geometry for sensing and PNT, and JCSAP extends both patterns to full three-function convergence. The survey emphasizes that waveform co-design, synchronization, interference mitigation, and resource management are the principal challenges once these services are unified, because each function imposes a different optimality criterion: rate and reliability for communications, ambiguity and resolution for sensing, and CRB/GDOP/PDoP for PNT (Sheemar et al., 30 Sep 2025).

6. Control, coordination, and autonomy

MFSS architectures become fragile unless control and autonomy are incorporated into the payload abstraction itself. Distributed-aperture beamforming is the clearest example. In a multi-satellite virtual antenna with $2304$7 to $2304$8 satellites and $2304$9 total radiating elements, Monte Carlo analysis shows that non-calibrated perturbations with translation standard deviation up to $330\ \mathrm{m}^2$0 and rotation standard deviation up to $330\ \mathrm{m}^2$1 can severely distort the pattern, while control-aware beamforming that uses actual position and attitude information can restore performance to near-nominal levels across the simulated range. To ensure with probability greater than $330\ \mathrm{m}^2$2 that mainlobe loss remains below 1 dB, HPBW-area change below 2%, and sidelobe-level increase below 1 dB in the non-calibrated case, position accuracy must be better than $330\ \mathrm{m}^2$3 and attitude accuracy must be within a few degrees; the paper also identifies radial separation and pitch/roll as the most sensitive perturbation components (Tuzi et al., 13 Oct 2025).

The same control–signal-processing coupling appears in laboratory and autonomy-oriented work. The Athena platform models multi-spacecraft phase-array communications using $330\ \mathrm{m}^2$4–$330\ \mathrm{m}^2$5 robotic agents, designed to scale to hundreds, with SDRs, air bearings, ducted fans, and onboard control. The underlying Artificial Neural Tissue framework is used to evolve distributed controllers that can self-organize formation, adapt to satellite loss or reassignment, and allow some satellites to be dynamically repurposed to split and communicate with multiple targets at once. This provides an explicit MFSS control motif: graceful degradation, dynamic task allocation, and phase-array coordination on a distributed architecture rather than on a single large relay spacecraft (Ravindran et al., 2017).

Autonomy at the constellation operations level is addressed in multi-agent reinforcement-learning studies for Earth observation. There, single-satellite planning is modeled as a POMDP and multi-satellite planning as a cooperative Dec-POMDP with actions Capture, Downlink, Charge, and Desaturate; the reward grants target priority for unique images, imposes a $330\ \mathrm{m}^2$6 penalty for failure, and forces learned trade-offs among imaging, energy, storage, and reaction-wheel limits. In realistic simulations with PPO, IPPO, MAPPO, and HAPPO, MARL is shown to balance imaging and resource management under partial observability and decentralized execution, with MAPPO and HAPPO providing stronger coordination than purely centralized or independent training in multi-satellite settings (Hady et al., 18 Jun 2025).

The control lesson is general. MFSS requires not only multi-function payloads but also multi-function decision loops: state estimation, beamforming, task assignment, and resource management must be closed around the same orbital and payload state variables.

7. Applications, limitations, and open directions

MFSS is already visible in application-driven architectures. Maritime surveillance is a particularly clear case because it naturally combines ship detection, tracking, environmental monitoring, pollution sensing, and security functions. A survey of space-based maritime surveillance identifies Sat-AIS, SAR, multispectral and hyperspectral optical payloads, and GNSS reflectometry as the main building blocks. It also highlights hosted and combined payload patterns, including SAR satellites carrying AIS receivers such as NovaSAR-1 and KOMPSAT-6, and small EO platforms such as the 3Cat series that mix AIS, GNSS-R, or hyperspectral functions. In MFSS terms, the same system can provide cooperative identification, non-cooperative detection, environmental monitoring, and sea-state estimation through coordinated use of heterogeneous payloads and constellations (Soldi et al., 2020).

At the same time, the literature is explicit about its simplifying assumptions. Space MIMO analyses often assume ideal or noiseless ISLs, perfect or compensated Doppler, flat fading, and single-user operation; distributed-aperture beamforming studies often assume perfect knowledge in the calibrated case and narrowband far-field conditions; ISAC formulations frequently use single-target or single-radar models; and navigation–communication integrations assume strong synchronization and tractable interference cancellation (Omid et al., 2023, Tuzi et al., 13 Oct 2025, Hou et al., 23 May 2026). These assumptions do not invalidate MFSS, but they define its current research frontier.

The open problems are correspondingly system-level. Surveys emphasize unified waveform design, statistical-CSI and delay–Doppler-robust beamforming, integrated channel models, constellation-level resource scheduling, geometry-aware orbit and handover design, interference management, synchronization, and secure multi-functional signaling (You et al., 2024, Sheemar et al., 30 Sep 2025). A plausible implication is that MFSS will be realized less as a single canonical architecture than as a family of co-designed infrastructures: some centered on distributed apertures and ISLs, some on unified waveforms, some on software-defined payloads, and some on autonomous coordination policies. What remains common across these variants is the replacement of function-specific spacecraft by shared, reconfigurable space systems whose dominant design variables are no longer only orbit and link budget, but also cross-functional coupling, calibration, and multi-objective optimization.