Stealthy Low Earth Orbit Satellite-To-Ground Quantum Communication

Quantum communication leverages fundamental principles of quantum mechanics to achieve information-theoretic security [17, 16]. At its core is quantum entanglement, where correlated quantum states enable parties to establish shared randomness. Quantum key distribution (QKD), exemplified by BB84 [3] and E91 [6], guarantees that any eavesdropping attempt introduces detectable disturbances due to the no-cloning theorem and measurement disturbance principle. As a result, legitimate parties can monitor the quantum bit error rate (QBER) to determine whether a key is secure.

While QKD has been successfully demonstrated over fiber-optic and free-space terrestrial links [4], these approaches face severe distance limitations. Optical fibers introduce exponential attenuation (0.2–0.3 dB/km), and free-space terrestrial QKD suffers from atmospheric scattering, turbulence, and alignment constraints. As a result, traditional ground-based QKD is generally limited to distances below 300 km.

LEO satellites offer a compelling alternative [1, 2, 15]. In satellite-based QKD, the photons traverse most of their path through vacuum, minimizing loss and allowing communication over thousands of kilometers [18, 13]. Experiments such as those conducted with China’s Micius satellite have demonstrated QKD over 1200+ km with key rates in the kilohertz range [11, 19].

LEO satellites orbit at altitudes of 500–2000 km and complete a revolution around the Earth every 90–120 minutes. These satellites periodically come into line-of-sight with ground stations, creating short but recurring time windows suitable for QKD. Each window typically lasts 100–500 seconds, during which entangled photon pairs are distributed to two ground stations.

Satellite-based QKD follows a procedure similar to terrestrial protocols but introduces new challenges due to the satellite’s mobility, Doppler shifts, beam pointing errors, and dynamic environmental conditions. Furthermore, satellite-ground links suffer from atmospheric attenuation near the Earth’s surface and must consider elevation angles to determine visibility and efficiency.

Existing research on QKD security has primarily focused on physical-layer vulnerabilities and countermeasures. Measurement-Device-Independent QKD (MDI-QKD) [17] addresses detection-side vulnerabilities, while device-independent QKD provides security even with untrusted devices. Quantum repeater networks [5, 12] aim to extend range through intermediate nodes, though practical implementation remains challenging.

However, these approaches largely neglect the system-level vulnerability introduced by satellite orbit predictability. Prior work on satellite QKD security has concentrated on improving physical transmission efficiency and reducing channel loss [7, 8], with limited attention to protecting the satellites themselves from being identified and targeted. The use of public TLE data for satellite tracking, while essential for coordination, creates an attack surface that has been largely overlooked in the quantum communication literature.

Recent work on satellite network security has addressed classical threats such as jamming and spoofing [21, 10], but these solutions do not account for the unique sensitivity of quantum channels to subtle disturbances caused by eavesdropping.

The convergence of predictable satellite orbits and the extreme sensitivity of quantum states to measurement creates a critical security gap. Unlike classical communication systems that can tolerate some level of interference, QKD links become unusable when QBER exceeds protocol-specific thresholds (typically 11% for E91). This makes quantum satellites particularly vulnerable to targeted attacks.

Our work addresses this gap by proposing the first system-level defense framework that leverages the scale and redundancy of modern LEO constellations. Unlike hardware-based solutions that require specialized quantum components, our approach operates at the network architecture level, making it complementary to existing physical-layer security measures. By combining stealthy deployment to prevent satellite identification with dynamic re-routing to maintain service continuity, we create a multi-layered defense that is uniquely suited to the challenges of satellite-based QKD.

The system model for satellite-based QKD uses LEO satellites as quantum sources to distribute entangled photon pairs for two ground stations, typically referred to as Alice and Bob. The process of QKD is shown in Figure 2. A full algebraic derivation is deferred to Appendix A.

Once the entangled photons are generated by the satellite, they are transmitted to the ground stations. After receiving the photons, both Alice and Bob perform measurements using randomly chosen bases, such as rectilinear or diagonal bases. Alice and Bob then engage in a process called basis comparison, where they publicly share the bases used for each measurement. They only keep the results where their measurement bases match. Alice and Bob then use the matched measurement results to generate shared secret keys. To ensure the security of the communication, Alice and Bob continuously monitor the QBER. If the error rate exceeds a predefined threshold, indicating a potential eavesdropping attempt, they will discard the affected key. Conversely, if no anomalies are detected, the key is accepted for secure use.

To reduce the risk of quantum satellites being identified and targeted, we camouflage their operational characteristics within a large LEO satellite network. This strategy includes the following components:

• $\mathrm{■}$

  Orbital and Visual Camouflage: Quantum-enabled satellites are assigned orbital parameters and physical profiles that match those of conventional communication satellites. This makes them indistinguishable from other nodes in the constellation when observed via standard tracking or optical means.

• $\mathrm{■}$

  Functional Obfuscation: Each quantum satellite is equipped with conventional communication modules and performs standard relay or communication tasks. As a result, its activity profile mimics that of a non-quantum node, adding further ambiguity.

• $\mathrm{■}$

  Beam Overlap: Entanglement distribution beams are spatially and spectrally overlapped with classical communication traffic. This beam multiplexing conceals the quantum transmission within broader traffic patterns, complicating signal isolation by potential eavesdroppers.

These techniques collectively reduce the probability that an adversary can detect, track, or distinguish quantum-enabled nodes, thus preempting targeted attacks.

Despite stealthy deployment, high-frequency eavesdropping attempts may still compromise QKD sessions. To maintain operational continuity, we introduce a dynamic re-routing mechanism that detects anomalies (e.g., elevated QBER) and immediately switches the QKD session to an alternative quantum satellite.

• $\mathrm{■}$

  Redundant Path Planning: During the planning phase, each entanglement session is pre-associated with multiple candidate satellites that have overlapping visibility windows. This redundancy enables seamless failover if an attack is detected, and eliminates real-time path calculation overhead.

• $\mathrm{■}$

  Responsive Channel Switching: Upon detection of anomalous QBER values, Alice and Bob notify the network to suspend the current session. The QKD task is then re-assigned to a backup satellite with an active line-of-sight to both stations.

• $\mathrm{■}$

  Resilience to High-Frequency Eavesdropping: Because each attack only disrupts one session, and the system can immediately resume via another satellite, the adversary must compromise multiple satellites in rapid succession to significantly degrade key throughput.

Figure 4 and 5 illustrate the original and re-designed entanglement distribution processes to defend against detected eavesdropping. Different from the process in Figure 4, the new entanglement distribution process reroutes to a backup quantum source (LEO satellite), allowing a new quantum satellite to continue the QKD execution, thereby mitigating the risk of eavesdropping.

Figure 6 shows the dynamic re-routing process in a satellite-ground station entanglement distribution system. In step 1, the satellite in the top left distributes entangled photon pairs to two ground stations, Alice and Bob. Alice and Bob then measure the received photons and exchange information through the public channel in step 2. However, the QKD fails due to Eve’s eavesdropping between the satellite and Alice. After detecting the eavesdropping, Alice and Bob notify the satellite constellation and switch the QKD channel via dynamic re-routing. In the last step, the two ground stations generate the secure key based on photon pairs from the new quantum satellite.

Our space model is built upon real-world satellite data to ensure practical relevance and accuracy. We utilize publicly available Two-Line Element (TLE) data from the Starlink constellation, comprising approximately 6,000 operational satellites at 550 km altitude. The orbital dynamics are simulated using the Simplified General Perturbations 4 (SGP4) model, which provides kilometer-level position accuracy for short-term predictions.

The quantum satellite network consists of satellites $\text{Sat}=\{s_1,s_2,\mathrm{\dots },s_m\}$ and ground stations $\text{GS}=\{g_1,g_2,\mathrm{\dots },g_n\}$. Each satellite is characterized by standard orbital parameters:

where ${\theta }_i$ represents inclination (the tilt of the orbit relative to Earth’s equator), ${\mathrm{\Omega }}_i$ is the right ascension of the ascending node (horizontal orientation), $h_i$ is altitude above Earth’s surface, ${\omega }_i$ is the argument of perigee (point of closest approach to Earth), $M_i$ is the mean anomaly (position along orbit at reference time), $n_i$ is the mean motion (orbits per day), and $T_i$ is the epoch time (reference time for orbital parameters).

Ground stations are defined by geographic coordinates:

where ${\lambda }_j$ is latitude and ${\beta }_j$ is longitude.

The SGP4 model predicts satellite positions as:

with the complete constellation state described by:

Our analysis focuses on the current Starlink deployment of approximately 6,000 satellites as it represents a realistic operational scale while providing sufficient diversity for stealth deployment. The methodology naturally extends to future mega-constellations through straightforward parameter scaling.

Given the known coordinates of the satellite and the ground station, one can calculate the transmission beam range for entanglement distribution. The position of the satellite ${𝐫}_{\text{sat}}(t)$ and the ground station ${𝐫}_{\text{gs}}$ are used to determine the line-of-sight and the effective communication window. The visibility condition is given by:

Considering atmospheric losses, the transmission range of satellite-to-Earth station links is affected by the elevation angle. When considering the effects of the atmosphere, we ignore the effects caused by weather conditions and turbulence. Without considering the interference of background noise, the transmission efficiency $\eta$ is divided into two parts:

where ${\eta }_{atm}$ is determined by the atmosphere, and ${\eta }_{fs}$ is influenced by diffraction, following the same mathematical formulation as the atmospheric transmittance formula used in vacuum conditions. The atmosphere is modeled as an absorbing layer of equal thickness and uniformity everywhere. The quantum distribution link from satellite to ground station, where the distance transmitted in the atmosphere is determined by the zenith angle between the satellite and the ground station $\zeta$, is given by:

When the absolute value of $\zeta$ is greater than $\frac{\pi }{2}$, the distribution link intersects the ground plane, and in this case, the transmission efficiency is 0. The parameters $r$, $w_0$, $L_R$, and ${\eta }_{atm}^{\text{zen}}$ are used for the QKD network model.

In the vacuum of space, the photon transmittance from satellite to satellite can be represented as:

where $w(L)$ at the distance $L$ is defined as:

The transmission channel prediction involves calculating the instantaneous visibility condition, the resulting communication time window, and the transmission beam range while accounting for atmospheric losses. This comprehensive approach ensures that the transmission channel is well-defined and allows for precise targeting, although it also exposes the communication to potential eavesdropping.

At any given time $t$, a satellite $s$ is considered visible to a ground station $g$ if the elevation angle satisfies:

where $e(\cdot )$ denotes the elevation angle between the satellite and the ground station, and $e_{\min }$ is a predefined minimum elevation threshold.

A communication link $(s,g)$ is said to be available if this visibility condition holds continuously over a non-zero time interval. Accordingly, the communication time window $t_{\text{transmit}}$ is defined as the maximal contiguous interval:

such that $e({𝐫}_s(t),{𝐫}_g)\ge e_{\min }$ for all $t\in [t_1,t_2]$.

The diffraction angle $\theta$, which represents the transmission beam range, is the area within which the satellite’s communication signal can effectively reach the ground station. The angle $\theta$ can be derived using the diffraction limit formula, which relates the wavelength $\lambda$ of the transmitted signal and the initial aperture or beam width $D$:

The diffraction angle $\theta$ describes the extent of this spreading and thus directly influences the effective range of the transmission beam. This range can be calculated considering the satellite’s altitude $h$, the Earth’s curvature, and the beam divergence angle $\theta$:

The public nature of satellite transmission channels enables interception attempts that directly impact quantum bit error rates. We model Eve’s capabilities and their system-level effects through a comprehensive analysis of interception mechanics and error propagation.

Eve’s interception follows an intercept-resend strategy consisting of photon capture within the beam footprint using strategically positioned equipment, random basis measurement (X or Z basis with equal probability), and photon resending in the measured state to maintain channel appearance.

The quantum bit error rate quantifies eavesdropping impact and is defined as:

For the E91 protocol with ideal equipment, the baseline QBER is approximately 4.7% due to natural channel imperfections. Eve’s interception introduces additional errors because with probability 0.5, Eve chooses the wrong measurement basis, and each wrong basis measurement randomizes the resent photon state, creating a 0.25 probability of bit error per intercepted photon.

The resulting QBER under eavesdropping becomes:

where ${\text{QBER}}_{\text{base}}=0.047$ represents the baseline error rate from channel imperfections and $P_{\text{intercept}}$ denotes the fraction of intercepted photons.

To illustrate the basis mismatch impact, consider Alice sending $|+⟩$ in the X-basis. Eve intercepts and measures in the Z-basis (wrong choice with 50% probability), causing the measurement to collapse the state to $|0⟩$ or $|1⟩$ with equal probability. Eve then resends in the measured Z-basis state, and Bob measures in the original X-basis, obtaining an incorrect result with 50% probability. This single interception attempt thus has 25% probability of introducing a bit error, demonstrating the sensitivity of QKD to eavesdropping and validating our threat model.

In summary, Eve’s actions, although concealed, introduce detectable errors into the system. By carefully monitoring the error rate, Alice and Bob can identify potential eavesdropping attempts and ensure the security of their quantum communication.

The Satellite Data Center provides the foundational spatial-temporal context for quantum communication scheduling through precise orbital prediction and visibility analysis. This module continuously ingests publicly available TLE data from operational LEO constellations, with provenance tracking to ensure experimental reproducibility.

The orbit propagation submodule transforms TLE orbital elements into precise Earth-centered inertial coordinates using the standard SGP4 formulation:

where ${𝐫}_{\text{sat}}(t)$ denotes the satellite position vector at time $t$ derived from the TLE parameters. The propagation outputs latitude, longitude, altitude, and line-of-sight geometry at configurable temporal resolution.

The visibility and entanglement window estimation submodule computes, for each candidate ground-station pair $(g_a,g_b)$, the set of time intervals $W_{s,g_a,g_b}=\{[t_1,t_2]\}$ during which satellite $s$ simultaneously meets minimum elevation and link-quality thresholds for both stations. These windows represent the fundamental temporal constraints for entanglement distribution and are continuously updated as the constellation evolves.

The Quantum Simulator implements a comprehensive emulation of the E91 protocol with configurable physical parameters and adversarial conditions. This modular framework accurately models quantum state evolution, measurement processes, and environmental effects to provide realistic performance assessment under both ideal and contested operational environments.

The entangled photon generation module initializes quantum states according to the Bell state formulation:

This entangled state ensures perfect correlation between measurement outcomes when identical bases are employed. The simulator generates photon pairs in batches of up to ${10}^6$ entangled pairs per experimental run, reflecting practical satellite emission capabilities while maintaining computational efficiency.

Photon transmission modeling incorporates both free-space and atmospheric effects through a multi-stage efficiency calculation. The comprehensive link budget accounts for diffraction losses, atmospheric attenuation, and pointing errors:

where ${\eta }_{\text{fs}}$ represents free-space diffraction efficiency calculated using Gaussian beam propagation models, ${\eta }_{\text{atm}}$ models zenith-angle dependent atmospheric transmittance, ${\eta }_{\text{pointing}}$ accounts for satellite pointing inaccuracies, and ${\eta }_{\text{detection}}$ incorporates ground station receiver efficiency.

The measurement and basis selection module implements the core E91 protocol mechanics. Upon receiving photons, Alice and Bob each randomly select a measurement basis from the Z-basis (computational basis) and X-basis (diagonal basis). The Z-basis includes $|0⟩,|1⟩$, while the X-basis includes:

Measurements are simulated using matrix operators corresponding to the selected bases. If both parties use the same basis, their results are correlated and retained for key generation.

Key sifting and post-processing implement the classical communication phase of QKD. Following measurement, Alice and Bob publicly compare basis choices through an authenticated classical channel, retaining only outcomes where measurement bases aligned. The sifted key undergoes error correction using low-density parity-check codes with efficiency factor $f(Q)\approx 1.1$, followed by privacy amplification to eliminate potential eavesdropper information. The final secure key rate follows the established formulation:

where $S$ represents the raw key rate, $Q$ denotes the measured quantum bit error rate, and $H_2(Q)$ is the binary entropy function quantifying the information theoretically available to an eavesdropper.

The QBER monitoring and eavesdropping simulation module implements real-time security assessment by continuously tracking the quantum bit error rate across sifted key segments. The simulator incorporates configurable adversarial models, including intercept-resend attacks with basis selection strategies ranging from random choice to optimized interception. When the measured QBER exceeds the E91 security threshold of 11%, the system flags the key material as compromised and initiates countermeasures through the Dispatch Center.

The Dispatch Center operates as the strategic control layer that orchestrates quantum communication sessions and executes defense mechanisms in response to detected threats, bridging predictive capabilities with physical-layer emulation.

QKD session scheduling employs a utility-based optimization that ranks candidate entanglement windows by expected key yield:

where $\widehat{S}(w)$ and $\widehat{Q}(w)$ are predicted raw rate and QBER for the window, $C_{\text{switch}}$ captures operational switching costs, and $\alpha$ is a tunable trade-off parameter. The scheduler allocates sessions to maximize cumulative expected key yield subject to operational constraints.

The dynamic re-routing mechanism implements our primary active defense strategy by rapidly responding to security incidents. Upon detecting QBER exceeding the policy threshold $\tau$, the system immediately suspends the compromised session and executes a handover to a pre-identified backup satellite that maximizes residual utility within overlapping visibility constraints. To mitigate adversarial forcing through repeated triggerings, the policy incorporates rate-limiting and increasing switch penalties for consecutive failovers, favoring stability unless substantial utility gains justify switching.

Practical considerations for denial-of-service resistance include statistical confirmation of anomalies through multiple QBER samples before triggering costly switches, adaptive penalty mechanisms that increase with switching frequency, and optional randomized backup satellite selection to enhance adversarial uncertainty about quantum payload distribution.

The integrated operation of these three modules creates a resilient quantum communication infrastructure capable of adapting to both natural network dynamics and active adversarial interference. Through continuous monitoring, rapid response mechanisms, and strategic resource allocation, our system design provides the foundation for practical satellite-based QKD deployment in contested environments.

Our experimental evaluation employs a rigorous methodology grounded in real-world satellite data and established quantum communication parameters. We utilize TLE data from the Starlink constellation (Space-Track.org, January 2024) comprising 6,174 operational satellites at 550 km altitude across three orbital planes (53°, 70°, 98°). The quantum simulator implements the E91 protocol with parameters validated against Micius satellite experiments: wavelength $\lambda =800$ nm, transmitter aperture $D=0.2$ m, receiver aperture $r=0.5$ m, and baseline QBER = 4.7%. All experiments simulate one-month operational windows with photon pair emission rates of 10 Gb/s, incorporating realistic channel losses and atmospheric effects.

Our constellation-wide analysis reveals the fundamental scalability advantages of LEO satellite networks for global QKD. Through large-scale simulation of up to 50 randomly sampled Starlink satellites, we quantify the aggregate availability of secure entanglement links over 24-hour periods. Each satellite provides 8-10 viable QKD windows daily for fixed ground station pairs, with durations of 2-5 minutes. When scaled across multiple satellites, the constellation achieves near-continuous coverage, reducing key generation latency from hours to minutes.

The orbital diversity across inclination planes (53°, 70°, 98°) enables comprehensive global coverage, with mid-latitude satellites serving populated regions and polar orbits ensuring service availability in high-latitude areas. This spatial diversity forms the foundation for our defense strategies, providing the redundant paths essential for both stealthy deployment and dynamic re-routing.

Figure 8 illustrates the photon transmission behavior between one satellite and two ground stations located in New York and Washington, D.C., over a 5-minute window. Specifically, Figure 8(a) depicts the photon transmittance of the satellite distributing photons to each of the two ground stations. The efficiency rises as the satellite approaches each station and falls as it departs, following the elevation angle. Figure 8(b) captures the aggregate entangled photon pair transmission efficiency for both links. We define an entanglement distribution as effective if the link efficiency exceeds $1\times {10}^{-9}$. The region between the two dashed lines highlights the valid distribution window. Correspondingly, Figure 7 provides a geographic projection of the satellite’s trajectory during this interval. The satellite trajectory is indicated by the blue dashed line and the red solid line, with the ground stations located in New York and Washington, D.C. The red solid line denotes the orbital segment where the satellite can simultaneously reach both ground stations with sufficient fidelity.

We simulated the orbital trajectory of a satellite in January 2024 and calculated the entangled photon pairs transmission efficiency of the satellite to two target ground stations over the course of a month. Figure 9 shows the function of transmission efficiency over time. Due to the changing position of the LEO satellite relative to Earth coordinates, continuous quantum entanglement distribution is not possible. Instead, the effective entanglement distribution occurs within specific time windows along the timeline. Despite these intermittent windows, they reoccur consistently, ensuring that a single satellite can reliably distribute entangled photon pairs to both ground stations over long periods. Figure 10 presents the statistical distribution and probability density function (PDF) of transmission efficiency. The figure highlights efficiency values above 110 dB, which represent high-fidelity transmission opportunities suitable for entangled photon pair delivery.

We develop an entanglement distribution simulator that faithfully replicates the full QKD workflow based on the E91 protocol. The simulator encompasses four core stages: entangled photon pair generation, quantum state measurement, key sifting, and QBER estimation.

The qubit generator simulates the emission of $1\times {10}^6$ entangled photon pairs per experiment, which are independently transmitted to Alice and Bob. Both parties randomly select measurement bases (Z-basis or X-basis) for each received photon. During transmission, photon pairs are subject to channel noise, and potential interception by an adversary (Eve) is simulated by introducing basis-dependent measurement disturbances. When Eve measures a photon using a basis mismatched with Alice or Bob, the quantum state collapses, potentially altering the final measurement result.

The simulator also incorporates eavesdropping by an adversary (Eve) and noise models to investigate the performance of the entanglement distribution under different environmental conditions and the potential eavesdropping attack effects of Eve throughout the process. We set up different experimental scenarios to analyze the QBER of the entanglement distribution under different conditions. In each scenario, we perform ten experiments where the qubit generator produces $1\times {10}^6$ entangled photon pairs per experiment, which are then measured by Alice and Bob. They publicly exchange information to calculate the QBER for each experiment. Our experiment focuses on the QBER as metric for evaluating the effectiveness and security of QKD.

Figure 11 illustrates the QBER observed during the entanglement distribution between Alice and Bob, under signal-to-noise ratios (SNR) of -10 dB and 10 dB. The simulations are performed both in the presence and absence of an eavesdropper, Eve. Under high noise conditions (SNR = -10 dB), the QBER approaches 50% regardless of Eve’s presence, indicating that the generated keys are nearly random and therefore unusable. Conversely, under lower noise conditions (SNR = 10 dB), the QBER is 4.7% without Eve’s eavesdropping and increases to 27.5% with Eve’s eavesdropping. This increase is due to the fact that Eve’s measurements have a 50% probability of perturbing the entangled states, so that 50% of these altered measurements yield random 0 or 1 results, resulting in about 25% of the keys being invalid.

Figure 12 further explores the QBER variation with changing SNR levels during entanglement distribution. As SNR increases, QBER decreases in both scenarios (with and without Eve’s eavesdropping). However, in the presence of Eve, the QBER plateaus at around 25%, highlighting the threshold below which successful entanglement-based QKD cannot be achieved due to the potential for undetected eavesdropping. For secure key distribution, it is crucial that the QBER remains low, particularly when there is no eavesdropping, and that the disparity between QBER values in the presence and absence of eavesdropping is substantial, allowing for reliable detection of eavesdropping. The E91 protocol stipulates that keys are valid only if QBER is below 11%, necessitating low-noise conditions for effective entanglement distribution.

Moreover, the influence of randomly selecting measurement bases in contrast to the use of a fixed measurement basis within the E91 protocol was also investigated. Figure 13 demonstrates these options. When Alice and Bob randomly select their bases between the computational basis (Z-basis) and the diagonal basis (X-basis), the presence of an eavesdropper is significantly indicated by an increase in the QBER, which allows Alice and Bob to detect the eavesdropper. However, if Alice and Bob use a predetermined fixed basis (Z-basis or X-basis), and Eve is aware of this basis, there is no significant change in QBER, regardless of whether or not Eve is eavesdropping. This scenario demonstrates the potential security risk associated with the use of a fixed basis, which may prevent Alice and Bob’s ability to detect the eavesdropper and thereby compromise the inherent security of quantum communications.

To quantify the practical benefits of our defense strategies, we evaluate three deployment scenarios against increasingly sophisticated adversaries: Baseline, Stealthy Deployment, and Dynamic Re-Routing.

Our experimental results demonstrate that the combination of stealthy deployment and dynamic re-routing creates a layered defense architecture. Specifically, Stealthy Deployment offers a passive, low-cost enhancement that obscures quantum satellite identities. Dynamic Re-Routing adds a proactive, adaptive mechanism that addresses service continuity when stealth fails. This approach leverages the inherent properties of large-scale satellite constellations (such as spatial diversity and orbital predictability) to create security guarantees that scale with constellation size.

Moreover, our simulations indicate that equipping only a modest fraction of satellites within a large constellation with quantum capability can already yield meaningful redundancy for robust defense while maintaining cost efficiency. In this setting, the dominant additional cost arises from coordinating and scheduling these quantum nodes effectively, rather than from replicating specialized payloads across every satellite. This makes the approach amenable to incremental deployment and compatible with existing constellation architectures.

Most importantly, our strategies transform the fundamental security model of satellite QKD from vulnerable point-to-point links to resilient network-based security, addressing the critical system-level vulnerability that has been largely overlooked in previous work.