{"id":"0ecc1974-93b9-4325-af21-3f358e45c8d3","arxiv_id":"2608.12265","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"Closed-form formulas for coverage, connection number, and received power in heterogeneous LEO satellite networks modeled by a spherical Cox point process with random satellite ranges.","lead":"This paper builds a mathematical model of multiple LEO satellite fleets sharing coverage, where satellites sit on randomly placed orbits and have random communication ranges. It derives formulas for whether a user is covered, how many satellites cover them, and how much signal and interference they receive.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 5's Laplace transform of total received power (Eq. 17) omits the retention probability F_kappa(~K) in the inner integral; as written it treats every satellite in the angular window as retained, so the claimed closed-form total-power distribution is incorrect.","rationale":"I read the paper as a tractable geometric framework for infrastructure sharing, with the central mathematical content in Theorems 1-5. The Cox/Boolean construction and the PGFL derivations for connection probability, connection number, association distance, and the average total received power are self-consistent and are supported by the included Monte Carlo checks; I would not dispute those. My concern is narrower but precise: Theorem 5's Laplace transform is missing the random-range retention probability. The derivation in Appendix E passes through the thinned process, whose intensity measure includes F_kappa, but the displayed PGFL exponent drops that factor. This is an internal inconsistency, not a disagreement with an external modeling choice, so it is the most load-bearing issue I found. The reader's SIR concern is legitimate: non-covering transmitters would add interference in a real shared-spectrum deployment, and the abstract's closed-form SIR claim is not backed by a theorem. However, that is an explicitly stated and qualified modeling assumption, whereas Eq. (17) is a mathematical error as written. Because the rest of the framework is sound and the fix is local, I keep the reader's conditional-accept recommendation, so the verdict is unchanged.","tokens_in":23336,"tokens_out":12384,"duration_ms":112176,"concrete_test":"Recompute Eq. (17) from the probability generating functional of the marked Cox process: the inner integrand should be F_kappa(~K(phi,eta)) times (1 - L_H(s p / ~K(phi,eta)^alpha)), not (1 - L_H(...)). Then evaluate the displayed Eq. (17) and the corrected expression for the paper's own setup (lambda=60, mu=20, r_s=6921 km, c_M=916 km, alpha=2, exponential H) and compare both against a Monte Carlo estimate of L_T(s) that generates the Cox process, draws i.i.d. kappa and H, and retains only satellites with kappa greater than distance. If the corrected expression matches the simulation while Eq. (17) does not, the theorem is wrong as written.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section III-D/Theorem 5 claims a closed-form Laplace transform for total received power. Re-deriving from the marked Cox probability generating functional, the factor contributed by a satellite at distance K is 1 - F_kappa(K)(1 - L_H(s p K^-alpha)), where F_kappa is the survival function of the communication range. After taking logarithms, the inner integral must contain F_kappa(~K(phi,eta)) times (1 - L_H(s p / ~K^alpha)). The displayed Eq. (17) instead contains only (1 - L_H(...)), omitting F_kappa. This is not a modeling choice: the expectation in Eq. (16) and the other theorems all include F_kappa, and the retention indicator 1_{kappa > distance} appears in the definition of the thinned process Psi_p in Eq. (8). The omission sits inside an exponential, so it changes L_T(s) for every s>0 and makes the stated expression the Laplace transform of a process where every satellite in the angular window contributes regardless of its random communication range. Because the abstract advertises closed-form expressions for total received signal power, this is a load-bearing correctness gap. The SIR assumption flagged by the reader is also a limitation, but Eq. (17) is internally inconsistent and checkable directly.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops an analytical framework for satellite infrastructure sharing in heterogeneous LEO/MEO networks. It models orbital planes as a Poisson point process on a torus, places satellites on each orbit as a Poisson process, and assigns each satellite an i.i.d. communication range so that the coverage region is a spherical Cox-Boolean model. Using probability generating functional and Campbell arguments, Theorems 1–3 derive the connection probability, the Laplace transform of the connection number, and the mean connection number; Theorem 4 derives the association distance distribution; Theorem 5 gives the mean total received power and a claimed Laplace transform of total received power. The paper also presents Monte Carlo SIR CCDFs in Section III-E, a multi-altitude extension in Section IV-A, and a comparison with a down-sampled Starlink-like constellation in Section IV-B.","tokens_in":23598,"tokens_out":4557,"duration_ms":40027,"significance":"If the derivations are correct, the framework is a valuable addition to stochastic geometry for non-terrestrial networks: it is one of the first to combine orbit-structured Cox point processes with random spherical footprints, and Theorems 1–4 provide tractable, parameter-explicit expressions that show how orbital-plane density, per-orbit satellite density, and communication range jointly govern coverage and connectivity. The derivations follow standard PGFL and Campbell arguments and appear internally consistent for Theorems 1–4. The paper also makes an honest effort to validate against system-level simulation and a realistic constellation model. However, the advertised closed-form total-power Laplace transform in Eq. (17) is missing a retention-probability factor, and the SIR contribution is simulation-only despite being advertised as closed-form; both issues must be corrected before the claims are sound.","major_comments":[{"comment":"The Laplace transform of the total received power omits the retention probability \\bar F_κ(\\tilde K(φ,η)) in the inner integral. Re-deriving from the conditional PGFL of the thinned satellite process ψ_i^p, the inner integral must be ∫_0^{\\tilde ω(φ)} \\bar F_κ(\\tilde K(φ,η)) (1 - L_H(s \\bar p / \\tilde K^α(φ,η))) dη; the displayed equation contains only (1 - L_H(...)). This is internally inconsistent with Eq. (16), where \\bar F_κ appears, and with the definition of Ψ_p in Eq. (8). Because the missing factor sits inside an exponential, it changes L_T(s) for every s>0 and invalidates the stated closed-form total-power Laplace transform, which is one of the paper's advertised contributions.","section":"Section III-D, Eq. (17)"},{"comment":"The abstract advertises 'closed-form expressions ... including ... the signal-to-interference ratio (SIR) distribution', but Section III-E contains no analytical SIR derivation; Figs. 7 and 8 report Monte Carlo CCDFs only. In addition, Eq. (19) defines interferers as exactly the co-covering satellites Ψ_p \\ X^*, so any satellite outside the typical user's coverage cells that still transmits on the shared spectrum is excluded from the interference sum. This assumption is load-bearing for the SIR curves and should be stated as a restrictive modeling assumption or relaxed, and the abstract should be corrected to describe the SIR results as simulation-based.","section":"Abstract and Section III-E"},{"comment":"The Cox model parameters λ and μ are calibrated to the average number of visible orbits and visible satellites measured from the same Starlink-like constellation that is then used for comparison of the mean connection number. The agreement in Fig. 10 is therefore partly enforced by the calibration step, not an independent test of the orbit-structure hypothesis. The paper acknowledges this, but to support the 'practical relevance' claim it should show sensitivity of the comparison to the calibration (e.g., uncalibrated parameters or a second constellation geometry).","section":"Section IV-B, Fig. 10"}],"minor_comments":[{"comment":"The SIR results should be described as simulation-based; the text currently says 'we evaluate the SIR distribution' without a theorem, which is easy to misread as analytic.","section":"Section III-E"},{"comment":"Eq. (33) has an extra closing parenthesis in arcsin(\\sqrt{1 - cos^2(\\hat φ(z)) sec^2(φ)}))); this typo should be corrected.","section":"Appendix D, Eq. (33)"},{"comment":"The notation table lists F_κ but not \\bar F_κ(x) = P(κ > x), even though the CCDF is used throughout; add it for completeness.","section":"Table I"},{"comment":"The connection number is called 'a nonnegative discrete random variable' in Section II-E2 and later 'a positive random variable' in Section III-B; clarify whether it can be zero.","section":"Section II-E"}],"recommendation":"major_revision","confidential_remarks":"The paper is within the scope of the journal, and the central Cox-PGFL framework is sound. The main risks are the Eq. (17) omission and the SIR overclaim; both are fixable in revision. No concerns about citation patterns or novelty disclosure."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear [colleague],\n\nI read arXiv:2608.12265. Short version: the geometric core is solid and worth engaging, but there is a definite error in the Laplace transform of total received power, and the abstract is stretched on the SIR claim.\n\nWhat's new: the author takes the spherical Cox model of orbits and adds i.i.d. communication ranges to form a Cox-Boolean coverage model, then derives the connection probability, connection number, nearest association distance, and mean total received power. Theorems 1-3 and the distance CDF are standard PGFL algebra and appear correct; the simulation checks in Figs. 3, 5, and 6 match. The multi-altitude extension (Prop. 1) is straightforward but a useful addition.\n\nSoft spots:\n\n1. Theorem 5's Laplace transform in Eq. (17) is missing the retention probability. For a satellite at distance K, the PGFL factor is 1 - F_bar_kappa(K)(1 - L_H(s p_bar / K^alpha)), not 1 - L_H(...). The displayed expression corresponds to a process where every satellite in the angular window contributes regardless of its random range, which contradicts Eq. (8) and even Eq. (16) in the same theorem. This is not a modeling choice; it changes L_T(s) for all s and invalidates the claimed closed-form for the total received power distribution. The mean E[T] in Eq. (16) is fine.\n\n2. The abstract claims \"closed-form expressions... for the SIR distribution,\" but Section III-E only provides simulated CCDFs. The text itself says \"we evaluate the SIR distribution... the empirical CCDF is obtained over multiple realizations.\" So that claim needs to be scaled back to a numerical study. Related: the interference model only counts co-covering satellites. If non-covering satellites transmit on the shared spectrum, the interference sum in Eq. (19) is incomplete and the SIR curves are optimistic. That may be a defensible baseline, but it should be stated as an assumption rather than implied to be the general SIR.\n\n3. The Starlink comparison calibrates lambda and mu to visibility statistics from the same constellation and then compares the mean connection number. The paper is transparent about this, and the match is a first-order sanity check, not independent validation. Minor.\n\nThe citation pattern is mostly the author's own prior work on the Cox model plus standard references; that's fair given the model lineage.\n\nWho's this for: researchers in satellite stochastic geometry and NTN dimensioning who want a tractable treatment of orbit-induced coverage overlap. The paper deserves serious refereeing: the fix for Eq. (17) is mechanical but necessary, and once corrected, the framework will be more useful.\n\nMy recommendation: engage with it—send to review—but make the SIR/closed-form language and Theorem 5 a condition for acceptance.","headline":"Solid Cox-Boolean geometry for LEO coverage overlap, but Theorem 5's Laplace transform drops the retention probability and the abstract overclaims a closed-form SIR result.","tokens_in":24101,"tokens_out":4268,"would_cite":true,"duration_ms":33639,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["60D05","60G55"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper argues that uncoordinated multi-operator LEO satellite sharing can be modeled in closed form by a spherical Cox-Boolean process, yielding explicit formulas for connection probability, connection number, association distance…","keywords":["satellite infrastructure sharing","stochastic geometry","Cox point process","Boolean model","LEO constellations","connection probability","connection number","signal-to-interference ratio"],"falsifier":"In the regime λ=60, μ=20, c_M=900 km, compare the SIR CCDF from Eq. (19) with a simulation that includes as interferers all satellites within distance c_M of the user—not just those whose random range covers the user—and check whether the reported P(SIR>τ) curves shift downward; a material downward shift would show the model's SIR claim is optimistic.","tokens_in":23120,"feed_emoji":"🛰️","tokens_out":7281,"duration_ms":63759,"temperature":0.7,"pith_summary":"The paper argues that when multiple LEO operators share satellite infrastructure without coordinating their orbits, the resulting coverage of the Earth is statistically tractable: the joint distribution of orbits and satellites can be described by a spherical Cox point process, and the union of circular coverage footprints forms a spherical Cox-Boolean model. On this basis the paper derives closed-form expressions for the probability that a typical user is covered, the distribution and mean of the number of satellites covering that user, the distance to the nearest covering satellite, and the total received power, with an empirical SIR characterization under nearest-association. These formulas make the trade-offs of sharing explicit: combining operators roughly adds their coverage linearly, but enlarging communication ranges or adding orbital planes also multiplies the interferers that share the spectrum, so geometric coverage gains do not automatically translate into better link quality. A comparison with a realistic inclined multi-shell constellation indicates the Cox model captures the scale and trend of coverage overlap after calibrating its parameters to visibility statistics. If correct, the framework lets system designers quantify sharing benefits without Monte Carlo simulation.","feed_headline":"Formulas predict coverage from pooled LEO constellations","feed_subtitle":"Exact formulas cover connection, redundancy, and power when operators pool uncoordinated orbits.","key_machinery":"The central object is the spherical Cox-Boolean model: a Cox point process on a sphere of radius r_s generates orbital planes (a Poisson process of isotropic orbit orientations) and then places Poisson-distributed satellites along each orbit; each satellite carries an i.i.d. mark κ, the communication range, so its footprint on Earth is a spherical cap C_{i,j}, and the union of these caps is a Boolean model driven by the Cox process. The analysis works by conditioning on the orbit process, using the probability generating functional of the one-dimensional Poisson satellite process per orbit to obtain void probabilities and Laplace transforms, and then applying the probability generating functional of the orbit process; all final expressions are single or double integrals over the angle window in which an orbit can come within the maximum range of the typical user. This machinery converts a seemingly unstructured multi-operator deployment into explicit formulas whose only inputs are λ, μ, and the complementary CDF of the communication range.","core_discovery":"The central claim is that the geometric and communication performance of shared, uncoordinated satellite networks is governed by three parameters—the mean number of orbital planes λ, the mean number of satellites per orbit μ, and the distribution of random communication ranges κ—and can be expressed in closed form. The proof strategy conditions on the orbit process, then on the satellite phases along each orbit, and applies the probability generating functionals of the two nested Poisson processes; the resulting formulas for connection probability (Theorem 1), connection number Laplace transform and mean (Theorems 2 and 3), association distance (Theorem 4), and total received power (Theorem 5) all reduce to integrals over an angular wedge defined by the maximum range c_M. The paper also shows the mean connection number scales linearly in λ and μ, that multi-altitude operators factor into a product of outage probabilities (Proposition 1), and that a calibrated Cox model approximates the mean connection number of a down-sampled realistic LEO constellation. A baseline SIR analysis, with the serving satellite as the nearest co-covering satellite and all other co-covering satellites as interferers, indicates that adding orbital planes helps at moderate SIR thresholds but saturates, while increasing c_M degrades SIR because overlap interference grows faster than the desired signal.","pith_inferences":["Beyond the paper, the explicit dependence of connection probability and total received power on λ, μ, and the range distribution invites a direct optimization: one could solve for the maximum c_M or constellation density that achieves a target SIR, producing a Pareto frontier of coverage versus interference.","The small but consistent overestimate of connection number in the realistic-constellation comparison suggests a latitude-dependent correction could be derived by conditioning on the actual inclination distribution rather than full isotropy, which would sharpen the model for real inclined shells.","Because all final formulas depend on the range distribution only through its complementary CDF, the framework can be used to design the distribution of communication ranges—for example concentrating ranges near a moderate value—to shift the coverage-interference trade-off without changing constellation size.","The linear scaling of mean connection number with the product λμ could inform spectrum sharing policy: regulators might bound aggregate orbital density across operators rather than per-operator constellation size when multiple fleets share the same spectrum."],"forward_implications":["The mean connection number under infrastructure sharing is the arithmetic sum of the individual operators' mean connection numbers, so pooling three comparable constellations triples the expected number of satellites covering a user.","Connection probability rises through an exponentiated integral of orbit and satellite densities, meaning several sparse uncoordinated operators together can approach universal coverage even when each alone covers only a small fraction of users.","Increasing the maximum communication range improves geometric coverage but worsens SIR, because the number of co-covering interferers grows with the footprint; coverage-centric metrics alone are therefore misleading under shared spectrum.","Association distance shrinks as more orbital planes join the shared network, improving the geometry of the desired link even while aggregate interference grows.","For multi-altitude deployments, the connection probability factorizes as one minus the product of per-altitude outage probabilities, so each altitude shell can be evaluated independently and then combined."],"supporting_citations":[{"why":"Supplies the spherical Cox point process that generates orbital planes and then satellites along them, the foundation of the spatial model.","marker":"[39]"},{"why":"Establishes the time-invariance of the Cox satellite process and provides the downlink analysis techniques reused here.","marker":"[40]"},{"why":"Provides the analytical Cox model for LEO and MEO networks that this paper extends to coverage cells and infrastructure sharing.","marker":"[41]"},{"why":"Supplies the probability generating functional, Boolean model formalism, and Campbell formula used in the theorem proofs.","marker":"[16]"},{"why":"Provides the satellite altitude, transmit power, antenna gain, and wavelength parameters used in the numerical and simulation results.","marker":"[46]"},{"why":"Defines the Walker constellation structure used to build the down-sampled realistic multi-shell comparison constellation.","marker":"[48]"},{"why":"Basis for the inclination and shell structure of the realistic constellation used in the connection-number validation.","marker":"[9]"},{"why":"Represents the earlier spatially independent binomial/PPP satellite models that the Cox model is contrasted against for lacking orbital structure.","marker":"[25]"},{"why":"Recent Cox-process downlink SIR analysis under general fading that the paper's baseline SIR treatment builds on.","marker":"[44]"}],"fun_headline_variants":["Closed-form odds for shared LEO satellite coverage","Three parameters forecast pooled LEO network performance","Exact formulas for shared LEO constellation connectivity","Shared LEO coverage: exact connection and SIR formulas"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The SIR analysis assumes the only interferers are the other satellites whose coverage cells contain the typical user, so satellites that transmit on the same spectrum but fall outside the user's coverage footprint contribute no interference.","fun_headline_variants_meta":{"raw":{"variants":["Closed-form odds for shared LEO satellite coverage","Three parameters forecast pooled LEO network performance","Exact formulas for shared LEO constellation connectivity","Shared LEO coverage: exact connection and SIR formulas"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000625,"raw_usage":{"total_tokens":2928,"prompt_tokens":1015,"completion_tokens":1913,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":631,"completion_tokens_details":{"reasoning_tokens":1853}},"tokens_in":631,"tokens_out":1913,"duration_ms":12841,"temperature":1.0,"reasoning_tokens":1853,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-16T00:10:53.521458+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"In the regime λ=60, μ=20, c_M=900 km, compare the SIR CCDF from Eq. (19) with a simulation that includes as interferers all satellites within distance c_M of the user—not just those whose random range covers the user—and check whether the reported P(SIR>τ) curves shift downward; a material downward shift would show the model's SIR claim is optimistic.","supporting_citations":[{"cited_title":"Cox point processes for multi altitude LEO satellite networks,","cited_arxiv_id":null,"evidence_quote":"Supplies the spherical Cox point process that generates orbital planes and then satellites along them, the foundation of the spatial model."},{"cited_title":"Modeling and analysis of downlink communications in a heterogeneous LEO satellite network,","cited_arxiv_id":null,"evidence_quote":"Establishes the time-invariance of the Cox satellite process and provides the downlink analysis techniques reused here."},{"cited_title":"A novel analytical model for LEO and MEO satellite networks based on Cox point processes,","cited_arxiv_id":null,"evidence_quote":"Provides the analytical Cox model for LEO and MEO networks that this paper extends to coverage cells and infrastructure sharing."},{"cited_title":"Stochastic geometry and wireless networks: volume I theory,","cited_arxiv_id":null,"evidence_quote":"Supplies the probability generating functional, Boolean model formalism, and Campbell formula used in the theorem proofs."},{"cited_title":"Solutions for NR to support non-terrestrial networks (NTN),","cited_arxiv_id":null,"evidence_quote":"Provides the satellite altitude, transmit power, antenna gain, and wavelength parameters used in the numerical and simulation results."},{"cited_title":"Satellite constellations,","cited_arxiv_id":null,"evidence_quote":"Defines the Walker constellation structure used to build the down-sampled realistic multi-shell comparison constellation."},{"cited_title":"SpaceX authorized for SCS and operations at lower altitudes","cited_arxiv_id":null,"evidence_quote":"Basis for the inclination and shell structure of the realistic constellation used in the connection-number validation."},{"cited_title":"Downlink coverage and rate analysis of low Earth orbit satellite con- stellations using stochastic geometry,","cited_arxiv_id":null,"evidence_quote":"Represents the earlier spatially independent binomial/PPP satellite models that the Cox model is contrasted against for lacking orbital structure."},{"cited_title":"Modeling and analysis for multiple-layer LEO satellite internet of things constellations,","cited_arxiv_id":null,"evidence_quote":"Recent Cox-process downlink SIR analysis under general fading that the paper's baseline SIR treatment builds on."}],"review_version":1}