{"id":"fdcf4bbf-e313-459d-858a-0bde5cf02fac","arxiv_id":"2602.21162","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"Phase-aware localization in pinching antenna systems achieves sub-meter accuracy in simulation, with a closed-form Cramér-Rao bound showing phase information scales as distance^-4 versus amplitude's distance^-6.","lead":"This paper derives the theoretical accuracy limit and a maximum-likelihood estimator for locating a user from signals received through a 'pinching antenna' waveguide, using both signal strength and phase, not just strength. It reports sub-meter positioning in simulation and explains why phase information helps more than amplitude at typical distances.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Phase-coherence assumption in eqs. (3)–(5) is load-bearing: without it the d^-4 phase information vanishes and the claimed advantage over amplitude-only WLS collapses.","rationale":"The reader's weakest_assumption identifies the same load-bearing concern. The derivation in §III is internally consistent under the stated model: the FIM/CRLB algebra is standard, and the d^-4 vs d^-6 scaling reflects the relative contribution of the phase and amplitude derivatives. I do not see a mathematical error in the model or the CRLB. However, the practical central claim depends on exact phase coherence across sequentially activated PAs, which is an untested and unflagged assumption. The phase information that produces the d^-4 term is precisely the information lost if per-slot phase offsets are unknown and independent. Since the numerical experiments inherit the ideal phase-coherent model, they cannot demonstrate robustness of the claimed advantage. This warrants a conditional verdict, not acceptance at face value. The proposed concrete test would directly settle whether the concern lands. I therefore keep the reader's CONDITIONAL verdict unchanged.","tokens_in":7043,"tokens_out":7998,"duration_ms":89087,"concrete_test":"Re-run the Fig. 1 comparison with the generative model augmented by per-slot unknown phase offsets: r_{k,n} = e^{jθ_n} s_{k,n}(u_k) + w_{k,n}, with θ_n ~ Uniform[0, 2π), and estimate u_k after marginalizing or jointly estimating the θ_n, using the same noise powers and N values as in Fig. 1. If the proposed estimator no longer consistently outperforms the amplitude-only WLS benchmark, the phase-coherence assumption is load-bearing.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The signal model in eq. (5) and the likelihood in eq. (7) contain no unknown carrier-phase term. Each received sample is modeled as the exact complex value s_k(u_k) plus AWGN, so the FIM in §III and the ML estimator in §IV both condition on perfect phase coherence across the N sequentially activated pinching antennas. The central claim—that phase-aware processing beats amplitude-only WLS—rests entirely on the e^{-j2πd/λ} term carrying distance information. In a sequential PA scan, each measurement is a separate time slot; independent oscillator phase, switching transients, or carrier-frequency drift introduce per-slot phase offsets θ_n. With independent θ_n, the phase term in (5) is uninformative, the Fisher information reduces to the amplitude-only d^-6 term, and the d^-4 phase enhancement in §III disappears; the proposed estimator then degenerates toward the [16] benchmark. A common unknown offset would be less destructive because phase differences between PAs would still carry information, but it would still invalidate the absolute-phase FIM as written. The paper neither states nor stress-tests this coherence assumption, and the numerical results in §V use the same idealized model, so the reported sub-meter accuracy and the performance gap in Figs. 1–2 are conditional on it.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper addresses user localization in pinching antenna systems (PASS) by using both amplitude and phase of the received complex baseband signal. A signal model is introduced that includes free-space path loss, waveguide attenuation, and distance-dependent phase rotation. The Fisher information matrix (FIM) is derived, leading to closed-form CRLB and position error bound (PEB) expressions. The authors show that phase-induced Fisher information decays as d^{-4} while amplitude-induced information decays as d^{-6}, which they interpret as a fundamental advantage of phase-aware localization. A two-stage maximum likelihood estimator, combining coarse grid search and Levenberg-Marquardt refinement, is proposed. Numerical results compare the estimator with an amplitude-only weighted least squares benchmark and report lower positioning errors.","tokens_in":1249,"tokens_out":1199,"duration_ms":64048,"significance":"If the underlying assumptions hold, the paper makes a useful estimation-theoretic contribution to the emerging PASS localization literature. The CRLB/PEB derivations are self-contained and the algebra in Eqs. (9)-(22) is consistent with the stated complex-Gaussian model. The d^{-4} versus d^{-6} separation is a clean and potentially design-relevant insight. The proposed two-stage ML estimator is practical and the numerical study covers different noise powers, numbers of pinching antennas, and user locations. However, the practical significance is conditional on an unmodeled phase-coherence assumption and on the absence of phase-ambiguity effects; these are not stress-tested in the manuscript.","major_comments":[{"comment":"The signal model contains no unknown carrier-phase offset: each received sample is modeled as the exact complex value s_k(u_k) plus AWGN. In a sequential PA scan, each activation occurs in a separate time slot, so oscillator phase offsets, switching transients, or carrier drift can introduce per-slot or common phase terms. If the per-slot offsets are independent, the e^{-j2πd/λ} term carries no distance information and the Fisher information reduces to the amplitude-only d^{-6} term, collapsing the claimed phase advantage. If the offset is common but unknown, the absolute-phase FIM in Eqs. (15)-(19) is not the appropriate FIM. The paper neither states this coherence assumption nor provides a robustness analysis. This is load-bearing for the central claim.","section":"Section II, Eqs. (3)-(5); Section III"},{"comment":"The CRLB/PEB is a local bound. At 2.8 GHz, λ≈0.107 m and the user-antenna distances are several meters, so e^{-j2πd/λ} is highly periodic and the likelihood is multimodal. The reported PEB in Fig. 2(a) is 0.008-0.026 m while the proposed estimator achieves 0.5-4 m, a gap that can reflect phase ambiguity and local convergence rather than mere suboptimality. The paper does not discuss when the local CRLB is attainable or how phase wrapping affects the d^{-4} advantage at low SNR. This should be addressed for the PEB to be a valid design target.","section":"Section III, Eq. (22); Fig. 2"}],"minor_comments":[{"comment":"The text says results are averaged over 1000 independent realizations, but the Fig. 2 caption says each point is averaged over 100 trials. Please harmonize.","section":"Section V, Fig. 2 caption"},{"comment":"The transmit pilot symbol s_k is said to be unit-power, but it appears inside the received signal expression without a definition of its phase or modulation. If s_k is known, its phase can be absorbed into the channel; if unknown, this is another phase nuisance. Please clarify.","section":"Section II, Eq. (5)"},{"comment":"The grid spacing d_grid=λ/4 and the number of initial points N_u=20 are free parameters. A brief sensitivity study or a comment on their choice would help reproducibility.","section":"Section IV"},{"comment":"The benchmark comparison is algorithmic only. Adding the amplitude-only CRLB/PEB would directly validate the claimed d^{-4}/d^{-6} advantage and make the comparison more informative.","section":"Section V, Fig. 2"},{"comment":"Minor typos and notation inconsistencies exist, e.g., the use of 'P As' versus 'PAs' and the unexpanded notation in Eq. (15). A careful proofreading pass is recommended.","section":"Throughout"}],"recommendation":"major_revision","confidential_remarks":"The technical core is sound under an idealized coherent-phase model, but the absence of any phase-offset modeling is a substantial gap for a localization paper. The authors should either justify the coherence assumption with a concrete system argument (e.g., shared oscillator and calibration) or extend the model to include nuisance phases and re-derive/quantify the achievable advantage. Without this, the central claim remains conditional on an assumption the paper does not acknowledge. The phase-ambiguity issue should also be discussed for the CRLB to be practically meaningful."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this is a well-executed letter. The complex-baseband model in (5), the FIM chain-rule derivation, and the closed-form PEB all check out. The observation that phase Fisher information scales as d^-4 while amplitude scales as d^-6 is genuinely the right way to frame why phase helps, and it is new for PASS to my knowledge. The ML estimator is standard but the two-stage grid+LM implementation is sensible and the comparison against [16] is fair: same geometry, same parameters, amplitude-only WLS. I believe the math. The d^-4/d^-6 split follows directly from |m|^2 and is not fitted.\n\nThe soft spot is exactly where the reader put it: the model treats phase as a known, coherent function of distance for every sequentially activated PA. Equation (5) has no unknown carrier phase offset, no per-slot phase term, no oscillator drift. If each time slot carries an independent phase offset, the e^{-j2πd/λ} term in the likelihood is uninformative and the Fisher information collapses to the amplitude-only d^-6 term; the advantage over [16] goes away. A common unknown offset is less destructive but still changes the FIM as written. The paper doesn't state this as an assumption or test it, and the simulations inherit the same model, so 'sub-meter accuracy' is a model-level statement, not hardware-level. This is not a fatal flaw for a theory letter—CRLBs are always conditioned on a model—but it should be labeled clearly and ideally given a robustness section (e.g., independent phase offsets or unknown common phase). Otherwise readers will over-read the deployment claim.\n\nOther notes are minor. The claim that phase-sensitive information 'fully captures' geometry is a bit strong given the model ignores phase wrapping ambiguity—at 2.8 GHz and meter-scale distances, phase wraps are handled implicitly but not discussed. The paper also assumes γ=1 following [16], which is okay for comparison but worth flagging if this ever goes to hardware. The numerical section is adequate, 1000 trials, clear figures; the PEB gap to empirical error is unsurprising and not overclaimed.\n\nWho is this for? Researchers working on PASS-based sensing or 6G localization; they will use the PEB as a design target and the d^-4/d^-6 insight in future work. I would send it to review—it deserves a serious referee and probably a small revision asking for an explicit phase-coherence assumption and a robustness check. I would not desk reject.","headline":"Phase-aware PASS localization with correct CRLB algebra and a real d^-4 vs d^-6 insight, but the practical claim leans on an unmodelled phase-coherence assumption that should be surfaced before deployment talk.","tokens_in":7855,"tokens_out":1825,"would_cite":true,"duration_ms":17441,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Exploiting signal phase along with amplitude gives pinching-antenna localization a closed-form error bound and an estimator that beats amplitude-only approaches, because phase information decays as distance^-4 rather than distance^-6.","keywords":["pinching antenna systems","user localization","Cramér-Rao lower bound","position error bound","maximum likelihood estimation","phase-aware estimation","Fisher information","indoor positioning"],"falsifier":"Introduce independent random phase offsets (e.g., 10°–20°) at each antenna activation in a simulation or prototype of this PASS localization scheme. If the observed position error moves toward the amplitude-only benchmark and the empirical variance no longer matches the closed-form PEB, then the perfect-phase-coherence assumption is falsified.","tokens_in":6943,"feed_emoji":"📡","tokens_out":9008,"duration_ms":73644,"temperature":0.7,"pith_summary":"This paper tries to establish that user localization in pinching antenna systems (PASS) should use both amplitude and phase of the received signal, not just amplitude as earlier work did. The authors derive the Fisher information matrix and closed-form Cramér-Rao lower bound / position error bound, and show that phase-induced information decays as the fourth power of user-antenna distance whereas amplitude-induced information decays as the sixth power. That scaling is why phase-aware localization is intrinsically more accurate at meter-scale distances. They then build a two-stage maximum-likelihood estimator (coarse grid search plus damped least-squares refinement) and simulate it against an amplitude-only weighted least squares benchmark, reporting lower positioning error across noise powers, antenna counts, and locations, with sub-meter accuracy in the tested 6 m × 10 m area.","feed_headline":"Phase-aware localization beats amplitude-only in pinching antennas","feed_subtitle":"Phase information decays as distance^-4, amplitude as distance^-6, so sub-meter localization is achievable.","key_machinery":"The engine is the factorization of the Fisher information matrix, J = (2/σ²) Re{G^H M~ G}, where G is an N×2 matrix with rows [u_x, u_y - v_n] (geometric sensitivity) and M~ is a diagonal matrix of |m_{k,n}|², with m_{k,n} proportional to e^{-j2πd/λ}(1/d³ + j 2π/λ 1/d²). Squaring yields amplitude information decaying as d^-6 and phase information decaying as d^-4, which is the concrete scaling law behind the advantage. This factorization produces closed-form CRLB and PEB expressions, avoiding numerical derivatives. The estimator is a two-stage maximum-likelihood approach: a λ/4-spaced coarse grid search to initialize, then a damped nonlinear least-squares refinement, selecting the final loca","core_discovery":"The central claim is that the complex baseband signal from each pinching antenna—free-space path loss, waveguide attenuation, and distance-dependent phase rotation—carries distance information in both magnitude and phase, and a proper Fisher-information analysis shows the phase contribution dominates at range. The FIM factorizes into a geometric sensitivity matrix depending only on user coordinates relative to antenna positions and a diagonal distance-sensitivity matrix containing squared terms of the form 1/d^6 (from amplitude) plus (2π/λ)^2/d^4 (from phase), so the closed-form PEB explicitly separates geometry from distance sensitivity. The paper's two-stage ML estimator realizes this adva","pith_inferences":["Beyond the paper: the same phase-sensitivity structure could support joint channel estimation and beam alignment in PASS, since beamforming also depends on the distances encoded in the phase term.","Beyond the paper: the results implicitly assume perfect phase coherence across sequentially activated antennas; a practical deployment with unknown carrier phase offsets would need joint phase estimation or calibration, and without it the d^-4 advantage would shrink.","Beyond the paper: the FIM factorization is general enough to extend to 3D user localization or multi-waveguide PASS, where the geometric matrix changes but the distance-sensitivity diagonal structure remains.","Beyond the paper: at mmWave/THz frequencies the (2π/λ)² phase term grows relative to the 1/d^6 amplitude term, so phase-aware localization should become even more dominant—provided phase noise does not scale with carrier frequency."],"forward_implications":["Amplitude-only localization in PASS discards the distance information carried by phase and is therefore information-lossy whenever phase coherence can be maintained.","The closed-form PEB gives system designers a direct target for choosing the number and placement of pinching antennas and waveguide parameters to meet a positioning requirement without Monte Carlo simulation.","Sub-meter accuracy is demonstrated in a 6 m × 10 m indoor area with eight pinching antennas at moderate noise, per the paper's numerical results.","Because phase information decays more slowly than amplitude information, the advantage of phase-aware localization grows as the user moves farther from the antenna array.","The gap between the practical estimator's error and the theoretical PEB indicates that further algorithmic refinement could still reduce positioning error."],"fun_headline_variants":["Phase beats amplitude for pinching antenna localization","Pinching antennas: phase info decays slower, wins localization","Phase-aware localization outperforms amplitude-only in pinching arrays","Why phase-aware localization wins in pinching antenna systems","Phase info decays slower, boosting pinching antenna localization"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The load-bearing premise is that the phase rotation e^{-j2πd/λ} is known exactly for every activated pinching antenna—no unknown carrier phase offset, oscillator drift, or phase ambiguity between time slots—so the phase genuinely encodes distance; if that coherence fails, the phase no longer carries distance information and the claimed d^-4 advantage collapses.","fun_headline_variants_meta":{"raw":{"variants":["Phase beats amplitude for pinching antenna localization","Pinching antennas: phase info decays slower, wins localization","Phase-aware localization outperforms amplitude-only in pinching arrays","Why phase-aware localization wins in pinching antenna systems","Phase info decays slower, boosting pinching antenna localization"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000182,"raw_usage":{"total_tokens":1142,"prompt_tokens":730,"completion_tokens":412,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":474,"completion_tokens_details":{"reasoning_tokens":336}},"tokens_in":474,"tokens_out":412,"duration_ms":3988,"temperature":1.0,"reasoning_tokens":336,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-02T21:05:18.998735+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Introduce independent random phase offsets (e.g., 10°–20°) at each antenna activation in a simulation or prototype of this PASS localization scheme. If the observed position error moves toward the amplitude-only benchmark and the empirical variance no longer matches the closed-form PEB, then the perfect-phase-coherence assumption is falsified.","supporting_citations":[],"review_version":1}