{"id":"d1099f83-ef2b-4009-a42c-e61212f6fc45","arxiv_id":"2607.13485","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"Users can self-locate to centimeter accuracy from downlink timing and power measurements in a pinching-antenna corridor, jointly estimating unknown antenna positions via Lambert W and weighted least squares.","lead":"A navigation method lets a mobile user locate itself in a corridor from signals sent through a single pinching-antenna waveguide, without knowing where the antennas are. It computes both antenna and user positions from signal timing and power, reaching centimeter-level accuracy in simulations within a ~16 m range.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Internal inconsistency in propagation constants: as printed, Eqs. (6)-(8) and (14) make the LWF-PAP closed form (17) off by a factor ε_r; the central estimation step does not follow from the stated model.","rationale":"The paper's central claim is that a user can recover PA positions and its own coordinates from TOA and received power without prior PA-position knowledge. That recovery rests entirely on the closed-form LWF-PAP expressions in Theorems 1-2. If those expressions are not algebraically derived from the stated channel model, the rest of the framework is moot. I checked the derivation and found a concrete internal inconsistency: the printed definitions of λ_g and v_g do not match Eq. (8), and eliminating y from Eq. (14) yields a log-term coefficient α/√ε_r, not α√ε_r as used in Appendix A. Under the printed α, these differ by a factor ε_r, which is non-negligible (ε_r=2.08 in simulations). The same mismatch propagates into the breakpoint distance d0 used for branch selection and into the error analysis. This is not an external-consensus disagreement; it is a checkable algebraic consistency issue. The reader's branch-selection concern is real but secondary: it is an operational limitation with a straightforward fix (detect clamping and flag as out-of-range). The covariance omission in Eq. (28) is also secondary: it undermines the minimum-variance/PA-PDOP theorems but does not invalidate the estimator itself. Because the inconsistency is likely a typo (correcting λ_g to λ/√ε_r restores the derivation), the appropriate action is to require the authors to correct the constants and re-run the noiseless check, not to reject the idea outright. The verdict therefore remains CONDITIONAL.","tokens_in":18917,"tokens_out":19512,"duration_ms":181227,"concrete_test":"Run a noiseless numerical consistency check: choose a geometry (e.g., d=4.7 m, y=4 m, ε_r=2.08, tanδ=4e-4, f_c=15 GHz), compute L and T from Eqs. (5)-(8) using the printed definitions (λ_g=λ√ε_r, v_g=c√ε_r), then plug L,T into Theorem 1 and compare recovered d,y to the true values. Independently, eliminate y from Eq. (14) by substituting y=(cT−d)/√ε_r and compare the resulting log-term coefficient to α√ε_r vs α/√ε_r. If the recovered distance/y do not match the true values, the central equations (17)-(18) do not follow from the stated model.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The LWF-PAP closed-form solution is not consistent with the model as printed. Section II defines α = π tanδ / λ_g with λ_g = λ√ε_r (Eq. 6) and τ_g = y_n / v_g with v_g = c√ε_r (Eq. 7), yet Eq. (8) uses τ_g = √ε_r y_n / c, which corresponds to v_g = c/√ε_r. Eliminating y_n from Eq. (14) gives ln(d) − (α/√ε_r)d = (ln10/20)L − (α c/√ε_r)T + ln η, i.e., coefficient α/√ε_r = π tanδ/(λ ε_r) = 1/(d0 ε_r) under the printed α. Appendix A and Theorems 1-2 instead use the coefficient α√ε_r = π tanδ/λ = 1/d0 and produce d = −d0 W0(−e^ξ/d0). Thus Eqs. (17)-(18) are not the solution of the stated equations unless λ_g is silently changed to λ/√ε_r. This is not a branch-selection or noise issue: with noiseless T and L, the reconstructed d_nk and y_n are biased by a factor near ε_r. The centimeter-level simulations therefore appear to test a different model than the text specifies. The branch-clamping concern raised by the reader is a practical limitation with an obvious flagging fix; the covariance omission in Eq. (28) affects the optimality claim but not the basic estimator. The parameter inconsistency is more load-bearing because it questions whether the central decoupling exists as written.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes a user-side navigation framework for a single-waveguide pinching-antenna system (PASS) in which a mobile user estimates both the unknown positions of the radiating PAs along the waveguide and its own coordinates using only downlink TOA and received-power measurements. The first stage (LWF-PAP) derives closed-form expressions for the PA-user distance and the PA position via the principal branch of the Lambert W function, and the second stage (WLS-PAN) converts the nonlinear navigation equations into an overdetermined linear system with a diagonally weighted least-squares solution. The authors also define a PA-derived position dilution of precision metric. Simulations under the paper's channel model report centimeter-level PA and user positioning accuracy within the breakpoint distance d0.","tokens_in":1645,"tokens_out":1592,"duration_ms":169269,"significance":"If the derivation were self-consistent, the paper would be a meaningful contribution to PASS-based localization: it removes the usual assumption that PA positions are known, it gives a closed-form decoupling of the otherwise coupled in-waveguide and free-space propagation parameters, and it provides a DOP-like performance metric for the anchor-line geometry. The Lambert-W inversion idea is elegant and the WLS formulation is practically appealing. However, the central estimation formulas do not follow from the propagation model as printed, and the claimed optimality of the WLS weights is not established by the current error analysis. These issues are load-bearing for the paper's main claims and require correction before the results can be relied upon.","major_comments":[{"comment":"There is an internal inconsistency in the propagation model. Eq. (6) defines α=π tanδ/λ_g with λ_g=λ√ε_r, and Eq. (7) sets v_g=c√ε_r; Eq. (8) nevertheless uses τ_g=√ε_r y/c, i.e. v_g=c/√ε_r. Eliminating y from (14) with the printed values gives ln d − (α/√ε_r)d = (ln10/20)L − (α/√ε_r)cT + ln η, and since α/√ε_r=1/(d0 ε_r), the coefficient is 1/(d0 ε_r). In contrast, Appendix A and Theorem 1 use the coefficient 1/d0 and produce d=−d0 W0(−e^ξ/d0) with ξ=(ln10/20)L − cT/d0 + ln η. Thus Eqs. (17)-(18) are not the solution of the stated system (14). With noiseless T and L the LWF-PAP output is biased by an ε_r-dependent amount, and the simulations appear to test a different model than the text specifies. The manuscript should correct the definitions in Eq. (6)/(7) (likely λ_g=λ/√ε_r and v_g=c/√ε_r) and the coefficient in (A.1), then re-derive all subsequent formulas consistently.","section":"Eqs. (6)-(8), (14), (A.1)-(A.2), Theorems 1-2"},{"comment":"The variance expression for ŷ_nk is not a correct first-order variance. From (23), y_nk=(cT_nk+d0 W(ζ_nk))/√ε_r, so the total derivative with respect to T includes the direct c/√ε_r term as well as the W dependence; since ∂ζ/∂T=−cζ/d0, the T-component is c²(1−S_nk ζ_nk)²/ε_r σ_t², not c²(S_nk²ζ_nk²+1)/ε_r σ_t². Equation (28) omits the cross term −2c²S_nkζ_nk/ε_r σ_t². Moreover, d̂_nk and ŷ_nk are functions of the same raw T and P measurements, so their estimation errors are correlated; Appendix B's (B.1) adds σ_d² and σ_y² as if they were independent when computing Var(b_nk). The claimed optimal weight matrix in Theorem 3 therefore does not follow from the stated derivation, and the 'minimum-variance unbiased' claim needs either a corrected covariance calculation or a more restricted statement.","section":"§III-B, Eq. (28); Appendix B, Eq. (B.1)"},{"comment":"The clamping in (24) silently replaces an out-of-domain Lambert argument by −1/e or 0. Because the user does not know d_nk a priori, it cannot tell whether the validity condition d_nk≤d0 in Lemma 1 is satisfied; when it is violated, the algorithm produces a plausible but biased pseudorange and PA position and feeds them into the WLS stage without any flag. This is not merely a numerical detail: the operating range is an assumption, not a condition the user can verify. The algorithm should detect and signal out-of-range cases, or fall back to another mode, rather than silently clamp.","section":"Eq. (24), Remark 2"}],"minor_comments":[{"comment":"'Pass loss' should be 'path loss' in the text near Eq. (19) and in Algorithm 1.","section":"§III-A, Algorithm 1"},{"comment":"The simulations use the same ideal channel model and the same noise model as the derivation; a mismatched-model test or experimental data would strengthen the reported centimeter-level accuracy claim.","section":"§IV"},{"comment":"Noise-variance notation is inconsistent: σ_t,nk and σ_p,nk are introduced in the model but drop the nk indices in the error-propagation equations. Please keep the indices or define the simplified notation.","section":"Eqs. (25)-(28)"},{"comment":"The description of [17] is repeated twice with slightly different wording ('AP-side user positioning' vs. 'AP-side positioning accuracy'); please check whether both statements are accurate.","section":"Introduction"},{"comment":"Figure captions are very sparse and do not identify curves, colors, or line styles; readers cannot decode the parameter variations from the captions alone.","section":"Figures"}],"recommendation":"major_revision","confidential_remarks":"The propagation-constant inconsistency is likely fixable by correcting the definitions of λ_g and v_g, so I would not reject the paper outright. However, the covariance error in Eq. (28) and the correlated-error issue in Appendix B mean that the 'optimal WLS' claim needs real mathematical rework, not just typographical correction. I would send the manuscript back for major revision and ask for a consistent derivation from a clearly stated propagation model."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"First, what's actually new: this is the first PASS positioning paper I know of that doesn't assume the PA positions are known to the user. The idea of using TOA and received power together, then applying the Lambert W function to separate the in-waveguide and free-space distance contributions, is genuinely original. The WLS-PAN formulation and PA-PDOP are natural extensions, and the single-waveguide geometry is handled sensibly. If the math worked, this would be a real contribution to user-side PASS navigation.\n\nThe math doesn't work as printed. Eq. (7) says the in-waveguide group velocity is v_g = c√ε_r, so the in-waveguide delay is y/(c√ε_r). Eq. (8) instead uses a total delay of (√ε_r y + d)/c, which corresponds to v_g = c/√ε_r. That's a factor ε_r mismatch within the model itself. When you eliminate y from (14), you get a Lambert equation with coefficient α/√ε_r = 1/(d0 ε_r). Appendix A and Theorem 1 solve with coefficient 1/d0, which is what you'd get from α√ε_r. So the closed forms (17)-(18) are not the solution to the stated equations. The simulations run the algorithm based on (17)-(18), so they test a different model than the text describes. This isn't a branch-selection issue or a small typo; it runs through the entire derivation.\n\nOther issues: Eq. (28) omits the covariance between T and ζ, so the minimum-variance claim and the PA-PDOP bound are not established. The clamp in Eq. (24) silently substitutes a spurious value when the user is outside the breakpoint; that's a practical limitation with an easy fix. And the simulations are self-consistent Monte Carlo with no baseline comparison.\n\nBottom line: a good idea with a load-bearing inconsistency. I'd send it to review, but the authors need to correct the propagation-constant definitions, re-derive the Lambert solution, and re-run the simulations. As is, I wouldn't cite it.","headline":"Real idea, but the printed model and the Lambert-W solution disagree by a factor of ε_r — the paper needs major revision before it's credible.","tokens_in":19808,"tokens_out":8710,"would_cite":false,"duration_ms":74620,"reading_group":"maybe","serious_thinker":"no","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Users can self-navigate to centimeter accuracy using only downlink broadcasts from a pinching-antenna waveguide, with no knowledge of antenna positions.","keywords":["pinching-antenna systems","user-side navigation","Lambert W function","weighted least squares","time-of-arrival","received power","anchor-line positioning","PA-PDOP"],"falsifier":"Deploy the paper's default system (15 GHz carrier, 12 m waveguide, 8 PAs, d0 ≈ 15.9 m) in a hall with laser-tracked ground truth, and have a user self-locate on a grid spanning true PA-user distances from 2 m to 25 m. If user-position RMSE does not stay at a few centimeters for all grid points with true distance below d0, the central accuracy claim is falsified; and if the points beyond d0 still return low error, the breakpoint condition is not the binding limitation the paper claims.","tokens_in":18755,"feed_emoji":"📡","tokens_out":9745,"duration_ms":89122,"temperature":0.7,"pith_summary":"To navigate with pinching-antenna systems (PASS), a mobile user usually needs to know the positions of the antennas along the waveguide; this paper claims that requirement can be dropped. The core idea is to treat each PA-user link as carrying two coupled distance-dependent signals—one travelling along the waveguide and one in free space—and to invert them together with the Lambert W function. From measured time-of-arrival and received power, the user obtains closed-form estimates of each antenna's position on the waveguide and the antenna-user pseudorange, with no calibration. A weighted least-squares stage fuses these estimates into the user's own coordinates, and a corridor-boundary constraint resolves the mirror ambiguity inherent in collinear anchors. Simulations report centimeter-level accuracy for both antennas and users within the breakpoint distance d0 = λ/(π tanδ), the range where the Lambert principal branch is valid.","feed_headline":"Locate yourself to centimeters with no antenna map","feed_subtitle":"Downlink signals alone give a user centimeter accuracy without any antenna coordinates.","key_machinery":"The load-bearing object is the Lambert W function, a special function that solves t e^t = z; here its principal branch W0 inverts the equation ln(d) − (α/√εr)d = ξ that couples the PA-user distance d and the PA position y. The argument of W0 must lie in [−1/e, 0], and this is exactly equivalent to d ≤ d0 = λ/(π tanδ), called the breakpoint distance—so the whole method operates only inside that range. Around this inversion the paper builds an anchor-line navigation geometry: all PAs lie on one line, so each navigation equation is a circle in the plane, and a two-unknown linear system is formed with the auxiliary variable v_k = x² + y². The corridor boundary x ∈ [0, D] selects the true interse","core_discovery":"The paper's central claim: the transcendental system coupling in-waveguide loss and free-space path loss—which appears when a user tries to separate two distance-dependent components from one downlink signal—has a unique closed-form solution via the principal branch of the Lambert W function. Under the breakpoint condition d_nk ≤ d0 = λ/(π tanδ), the PA-user distance is d_nk = −d0 W0(−e^{ξ_nk}/d0) and the PA position is y_n = (c T_nk + d0 W0(·))/√εr. This lets a user solve navigation without knowing PA coordinates. The WLS-PAN stage linearizes the circle-intersection equations with an auxiliary variable v_k = x² + y², uses the corridor boundary to discard the mirror solution, and derives the","pith_inferences":["The paper's own future-work list suggests multi-waveguide configurations; the natural extension of the same auxiliary-variable trick would fuse circles from different waveguide lines, which could remove the corridor-boundary ambiguity and generalize the method to open spaces.","The user still needs the waveguide's material constants (εr, tanδ) to compute d0, so the 'no prior knowledge' claim concerns antenna positions, not the waveguide itself; a sensitivity test around a mis-specified d0 would be a natural stress test.","The clamping in the paper's Lambert-argument construction means that when noise pushes the argument outside [−1/e, 0], the algorithm silently returns a valid-looking pseudorange and PA position instead of flagging an invalid measurement; flagging and discarding such measurements could improve robustness at low signal-to-noise ratio."],"forward_implications":["A deployed PASS waveguide becomes an immediate self-navigation infrastructure: users can compute positions without waiting for or trusting a broadcast antenna map.","Within the breakpoint distance, both the PA positions along the waveguide and the user positions reach centimeter-level RMSE in simulation, with the PA-derived position dilution of precision (PA-PDOP) below 0.3.","Uniform PA deployment outperforms random deployment, and increasing the number of PAs improves accuracy with diminishing returns.","The usable coverage range is set by material choice: a lower dielectric dissipation factor tanδ and a longer carrier wavelength both enlarge d0 and therefore the reliable working area.","Beyond the breakpoint distance, errors grow sharply and the Lambert-branch assumption breaks down, so the scheme is confined to short-range, line-of-sight operation."],"fun_headline_variants":["Downlink-only self-navigation hits centimeter accuracy with pinching antennas","Pinching-antenna navigation: users locate themselves without antenna coordinates","Closed-form Lambert W solution enables antenna-free self-location","User-side navigation from pinching antennas: no map, cm precision"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"Every PA-user distance must satisfy d_nk ≤ d0 = λ/(π tanδ) so that the principal branch of the Lambert W function is the correct branch, and the user cannot verify this condition from its measurements; when it is violated, the algorithm clamps the Lambert argument and silently substitutes a spurious solution.","fun_headline_variants_meta":{"raw":{"variants":["Downlink-only self-navigation hits centimeter accuracy with pinching antennas","Pinching-antenna navigation: users locate themselves without antenna coordinates","Closed-form Lambert W solution enables antenna-free self-location","User-side navigation from pinching antennas: no map, cm precision"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000441,"raw_usage":{"total_tokens":2100,"prompt_tokens":797,"completion_tokens":1303,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":541,"completion_tokens_details":{"reasoning_tokens":1230}},"tokens_in":541,"tokens_out":1303,"duration_ms":13781,"temperature":1.0,"reasoning_tokens":1230,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-02T05:01:46.491078+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Deploy the paper's default system (15 GHz carrier, 12 m waveguide, 8 PAs, d0 ≈ 15.9 m) in a hall with laser-tracked ground truth, and have a user self-locate on a grid spanning true PA-user distances from 2 m to 25 m. If user-position RMSE does not stay at a few centimeters for all grid points with true distance below d0, the central accuracy claim is falsified; and if the points beyond d0 still return low error, the breakpoint condition is not the binding limitation the paper claims.","supporting_citations":[],"review_version":1}