{"id":"38d65a45-ba5f-46d7-baf5-e5e35e75cb00","arxiv_id":"2607.27903","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"Using multiple chirp rates across time frames makes delay-Doppler estimation well-posed for Rydberg atomic quantum receivers, with an edge-distribution design that minimizes the post-chirp matrix condition number.","lead":"This paper proposes a wireless sensing waveform, multi-chirp AFDM, that lets Rydberg atomic quantum receivers estimate target distance and speed at the same time. The design resolves an ambiguity in earlier single-chirp schemes and is claimed to improve simulated estimation accuracy by up to two orders of magnitude.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"MC-AFDM model ignores that echoes with delay exceeding the 80 ns frame are produced by a previous post-chirp; simulated ranges up to 1000 m violate the model's implicit delay constraint.","rationale":"The reader's weakest assumption is the first-order Taylor linearization of the Rydberg response (a physical/domain assumption inherited from Ref. [18]). I find a more load-bearing, internal inconsistency: the time-segmented MC-AFDM waveform makes the same-frame phase-difference formula (21) invalid whenever the target delay exceeds the frame duration, and the simulation parameters violate that constraint by orders of magnitude. This is not a matter of external consensus but of the model contradicting the claimed numerical scenario. The mathematical rank argument itself remains valid for delays shorter than a sub-frame, so the central idea is salvageable by choosing a larger N or restricting the scenario to short ranges. Because the paper's core contribution can be repaired and the reader's conditional verdict already requires substantial revision, I recommend keeping the verdict at CONDITIONAL rather than moving to REJECT. The agreement field is 'disagree' because the reader's stated weakest assumption is not the most load-bearing one.","tokens_in":21929,"tokens_out":15079,"duration_ms":162283,"concrete_test":"Simulate the true MC-AFDM frame sequence for a single target at R=100 m (τ≈667 ns), using the actual transmit frame index p' = p − floor(τ/T_frame) for the echo and the current frame p for the LO. Compute the instantaneous phase difference over one frame and fit it with both (a) the paper's model cos(ω^(p)t + φ) and (b) a quadratic-phase model. If the residual after fit (a) exceeds the OMP grid spacing or the estimated ω varies by more than one grid step within the frame, the model is invalid. Additionally, rerun the NRMSE experiment with R constrained so that τ < 80 ns (e.g., R ≤ 10 m) and compare to the paper's Fig. 5; a dramatic performance collapse at R=100 m versus R=10 m would confirm the concern.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The MC-AFDM measurement model in Sec. IV assumes that the received echo and the LO reference both use the same per-frame phase function φ_m^(p)(·) of the current post-chirp c̃1^(p). This is only valid when τ_k − τ_l is smaller than the frame duration T_frame = NΔt, so that t−τ_k and t−τ_l fall in the same post-chirp segment. With N=8 and Δt=10 ns, T_frame = 80 ns, while the simulation uses ranges R_k ∈ [10, 1000] m, i.e., delays τ_k ∈ [66.7 ns, 6.67 μs]. Almost all of these delays exceed T_frame. Under the actual frame sequence, the echo arriving during frame p was transmitted during frame p' = p − floor(τ_k/T_frame), which has a different post-chirp c̃1^(p') ≠ c̃1^(p). The true phase difference is φ^(p')(t−τ_k) − φ^(p)(t−τ_l) + ν_k t, not the same-frame expression in (21). This phase difference contains a term proportional to (c̃1^(p') − c̃1^(p))·t², so the beat signal is not a constant-frequency cosine. Consequently, the linear model (32), the full-rank/C1 argument, the LS estimator (53), and the CRLB (58)–(59) do not describe the simulated scenario. The claimed 'two orders of magnitude' NRMSE improvement is therefore unvalidated: the numerical experiments run the estimator on a model that the waveform itself does not produce for those ranges.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript proposes a multi-chirp AFDM (MC-AFDM) waveform for joint delay-Doppler estimation using Rydberg atomic quantum receivers (RAQRs). The authors model the Rydberg self-heterodyne output as a sum of cosines whose frequencies are affine functions of delay and Doppler. They observe that the standard single-chirp AFDM (SC-AFDM) yields the same frequency vector for all subcarriers, producing only K equations for 2K parameters, and propose using P distinct post-chirps across P frames to make the system overdetermined. They then provide a condition-number-based design rule, an OMP-based frequency estimator followed by least-squares delay-Doppler recovery, and a Cramér–Rao lower bound. Simulations compare SC-AFDM, dual-chirp AFDM, and uniform/edge chirp distributions.","tokens_in":22323,"tokens_out":17411,"duration_ms":183187,"significance":"The core idea—using post-chirp diversity to lift a rank deficiency in the frequency-only delay-Doppler inversion—is simple and plausible, and the paper gives transparent derivations for the post-chirp matrix and the CRLB. The rank-deficiency observation in Remark 1 is correct for the frequency-only model, and the CRLB derivation is internally consistent under its stated asymptotic approximations. If the simulation and the design theorem were fully valid, the paper would be a useful contribution to RAQR waveform design. However, as detailed below, the numerical validation is based on a model that does not describe the simulated delay regime, and the proof of the central design theorem is not valid as stated.","major_comments":[{"comment":"The MC-AFDM measurement model in Eqs. (30)–(32) assumes that, at time t in frame p, both the reference signal and the delayed target echo use the current post-chirp c̃_1^(p). This is valid only for delays smaller than the frame duration T_frame = NΔt = 80 ns. The simulation in §VI-A draws R_k ∈ [10, 1000] m, i.e. τ_k ∈ [66.7 ns, 6.67 μs], with τ_l ≈ 6.7 ns. Thus for essentially all targets the echo arriving during frame p was transmitted in frame p' = p − floor(τ_k/T_frame), which uses a different post-chirp c̃_1^(p') ≠ c̃_1^(p). The phase difference then contains a term (c̃_1^(p') − c̃_1^(p))t², so the beat signal is not a constant-frequency cosine. Consequently Eqs. (32), (49), (53), and the CRLB (58)–(59) do not describe the simulated scenario, and the reported two-order-of-magnitude gains are not validated. The authors should either restrict the simulation to frame-compatible delays","section":"§IV-A and §VI-A"},{"comment":"The proof of Theorem 1 asserts that minimizing D1/√D2 is equivalent to maximizing Var(c̃_1), but D1 = 4Σ(c̃^(p))² + P depends on both the variance and the mean of the chirp-rate vector. The claim is not true in general. For example, with P = 3 and chirp rates restricted to [1,2], the maximal-variance vector (1,2,2) gives κ(C1) ≈ 13.74, while the lower-variance vector (1,1.5,2) gives κ(C1) ≈ 12.99. Thus maximizing variance does not always minimize the condition number. The theorem and the resulting edge-distribution design rule need either a corrected proof under the actual constraints of Eq. (42), or a reformulation that accounts for the dependence of D1 on the chirp-rate values.","section":"§IV-C3 and Appendix A"},{"comment":"The claim that SC-AFDM 'precludes reliable estimation' is established only for the fluctuation-frequency vector in Eq. (29), not for the complete measurement model in Eq. (26). The phase term φ_{m,k} in Eq. (21) depends on τ_k − τ_l through the subcarrier-dependent term −m/(NΔt)(τ_k − τ_l), so delay information is in principle encoded in phase differences across chirp-subcarriers even with a single post-chirp. The rank-deficiency of C1 and the CRLB derivation in Appendix B treat φ as an independent nuisance parameter and therefore discard this known coupling. If the authors intend a frequency-only estimation strategy, they should state this explicitly and justify why phase information cannot resolve the ambiguity; otherwise the 'optical ambiguity' motivation is overstated.","section":"§III-C, Remark 1, and Eq. (21)"}],"minor_comments":[{"comment":"The definition ω_k^(p) = (1/N)Σ_m ω_{m,k}^{(p)} is unnecessary because Eq. (32) already makes ω_{m,k}^{(p)} independent of m. The averaging step in Algorithm 1 could be simplified to avoid implying that there are m-dependent fluctuations in the model.","section":"Eqs. (32) and (37)"},{"comment":"The horizontal axis is labeled 'normalized variance' but the normalization is not defined. Please specify the normalization, e.g. Var(c̃_1)/(c̃_1^(max) − c̃_1^(min))², so the reader can interpret the condition-number curve.","section":"Fig. 4"},{"comment":"The NRMSE in Eq. (63) normalizes by |ζ_k|². The simulation setup sets R_k ∈ [10,1000] m and v_k ∈ [50,300] m/s, so the normalization is stable, but this should be stated; otherwise the metric can behave erratically for parameters near zero.","section":"§VI-A"},{"comment":"Since targets and noise are randomly generated, confidence intervals or Monte-Carlo error bars would strengthen the comparison between MC-AFDM-UD and MC-AFDM-ED, especially in Figs. 5 and 6 where the performance gaps are large but no variance information is reported.","section":"§VI"}],"recommendation":"major_revision","confidential_remarks":"The frame-delay inconsistency is the most serious issue: the headline numerical claims are obtained by running the proposed estimator on a model that the MC-AFDM waveform does not actually produce for ranges above about 12 m. The condition-number theorem also needs correction. Both problems are substantive but appear fixable in principle: the simulation framework must either be restricted to sub-frame delays or replaced by a genuine multi-frame echo model, and the design criterion must be proved under the stated constraints rather than inferred from variance alone. I therefore recommend major revision rather than rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Hi,\n\nTwo things to know: the core construction is real, and the simulation model is broken for the advertised ranges. The paper proposes using multiple AFDM post-chirps to turn an underdetermined delay-Doppler estimation problem into an overdetermined one. That is legitimate, and the rank-deficiency argument (Remarks 1 and 2) is correct. The CRLB derivation in Appendix B is also internally consistent given the stated approximations.\n\nThe problem is in Section VI. The measurement model assumes that the received echo and the LO reference use the same per-frame phase function φ_m^(p)(t). That is only valid when the differential delay τ_k − τ_l is smaller than the AFDM symbol duration NΔt = 80 ns. The simulations use ranges from 10 m to 1000 m, i.e., delays up to 6.67 μs. For those delays, the echo arriving during frame p was transmitted during frame p' = p − floor(τ_k/T_frame), which has a different post-chirp. The true phase difference then contains a quadratic term proportional to (c̃_1^(p') − c̃_1^(p))·t², so the beat signal is not a constant-frequency cosine. The linear model in (32), the OMP dictionary, the LS estimator, and the CRLB do not describe that scenario. The claimed two-orders-of-magnitude improvement is therefore not demonstrated.\n\nThis is fixable. Restrict the simulation to differential delays within one frame, or extend the model to handle inter-frame delays explicitly. They should also explain why an 80-ns frame is realistic for targets at hundreds of meters when the maximum unambiguous delay without frame crossing is 12 m.\n\nSmaller soft spots: The proof of Theorem 1 (Appendix A) claims minimizing κ(C1) is equivalent to maximizing Var(c̃1), but D1 also depends on the mean and sum of squares; that step is not justified. Edge distribution may still be optimal, but the proof needs work. The 'quantum-induced advantage' claim is unsupported because no classical receiver baseline is tested. The linearized Rydberg model with strong-LO assumption is inherited from [18] and could be sensitive to chirps sweeping across resonance, though that is a domain assumption rather than a mathematical error.\n\nWorth refereeing: the multiple-chirp disambiguation idea is relevant to a growing niche, and the CRLB framework is reusable. But the simulation mismatch is load-bearing and needs to be resolved before publication.","headline":"The multiple-chirp idea is sound, but the simulation model breaks for the advertised ranges, so the headline improvement is not yet credible.","tokens_in":22822,"tokens_out":5068,"would_cite":true,"duration_ms":48481,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Sending two chirps lets Rydberg receivers locate moving targets","keywords":["Rydberg atomic quantum receivers","affine frequency division multiplexing","delay-Doppler estimation","multi-chirp waveform","post-chirp optimization","quantum wireless sensing","orthogonal matching pursuit","Cramer-Rao lower bound"],"falsifier":"A controlled experiment with a single moving reflector in an anechoic chamber: transmit MC-AFDM with two known post-chirps through a Rydberg receiver, record the output voltage, and compare the two measured fluctuation frequencies against equation (32) using independently measured delay and Doppler. If the extracted frequencies deviate from the predicted linear relation by more than the noise floor, or if the least-squares delay-Doppler estimate is biased, the linearized cosine model (15) is falsified.","tokens_in":21850,"feed_emoji":"📡","tokens_out":3145,"duration_ms":33157,"temperature":0.7,"pith_summary":"This paper shows that the optical readout of a Rydberg atomic quantum receiver collapses each target's delay and Doppler into a single fluctuation frequency, leaving the classical single-chirp AFDM waveform with too few measurements to separate multiple targets. The proposed multi-chirp AFDM transmits several distinct post-chirps across time frames, making the post-chirp matrix full-rank and the estimation problem over-determined. The paper further proves that pushing the chirp rates to the edges of the allowed range minimizes the matrix's condition number, improving accuracy, and presents an OMP-plus-least-squares algorithm whose simulated performance approaches the Cramer-Rao lower bound. If correct, this would give Rydberg receivers a waveform that supports high-resolution multi-target range and velocity sensing in mobile, doubly-dispersive channels.","feed_headline":"Sending two chirps lets Rydberg receivers locate moving targets","feed_subtitle":"A new multi-chirp AFDM waveform resolves the optical ambiguity that previously hid delay and Doppler from atomic sensors.","key_machinery":"The post-chirp matrix C1, whose p-th row is [2˜c_1^(p), -1], is the central object: it linearly maps each target's channel parameter vector [τ_k−τ_l, ν_k] to the vector of per-frame fluctuation frequencies. Its full-rank condition requires at least two distinct chirp rates (P≥2), and its condition number—minimized by maximizing the variance of the chirp-rate vector, i.e., pushing chirp rates to the edges of the allowed range—controls the amplification of frequency-estimation error in the least-squares delay-Doppler recovery. The CRLB derived from the corresponding Fisher information matrix scales inversely with this variance and with P.","core_discovery":"The central claim is that the optical ambiguity of Rydberg atomic receivers in doubly-dispersive channels can be removed by using multiple distinct post-chirps in AFDM. In the classical single-chirp scheme, each target produces one fluctuation frequency per frame, giving K equations for 2K unknowns (delay and Doppler). With P≥2 distinct chirp rates, the post-chirp matrix C1—whose rows are [2c~_1^(p), -1]—becomes full-rank, yielding KP equations. The paper also discovers that maximizing the variance of the chirp-rate vector, achieved by placing chirp rates at the minimum and maximum allowed values, minimizes the condition number of C1 and sharpens delay-Doppler estimates. The resulting MC-AFD","pith_inferences":["The rank-deficiency argument is generic: any sensing system that maps delay and Doppler into a single frequency measurement with a unique slope faces the same K-equations-for-2K-unknowns ambiguity, so the multi-chirp trick likely transfers to other self-heterodyne or autocorrelation-based receivers beyond Rydberg atoms.","Because the CRLB scales as 1/Var(c~_1) and 1/P, the fundamental accuracy limit is set by the receiver's instantaneous bandwidth (which bounds the chirp-rate spread) and the number of frames; waveforms with non-linear chirps or intra-frame chirp-rate variations might relax this trade-off.","The minimum chirp-rate difference required by the OMP grid (Remark 3) implies a direct coupling between grid resolution, target range, and bandwidth; a gridless frequency estimator could reduce the required chirp spread or the number of frames.","The entire framework hinges on the linearized cosine measurement model, so an experimental calibration of (15) across the full chirp bandwidth is the next logical step before deployment; if the linearization fails near resonance, the fluctuation-frequency extraction would need a different detector model."],"forward_implications":["Rydberg atomic receivers can jointly estimate delay and Doppler for multiple moving targets in doubly-dispersive channels, which single-chirp AFDM cannot do.","Edge-distributed post-chirp rates improve range and velocity estimation by up to roughly 2.5- to 4-fold over uniform distribution in multi-target scenarios.","Increasing the number of post-chirps P lowers both the CRLB and the condition number, so estimation accuracy continues to improve with more chirp diversity.","The proposed OMP-plus-LS algorithm reaches the derived CRLB across simulated SNR levels, indicating the waveform, not the estimator, is the limiting factor.","The MC-AFDM design can be extended to integrated sensing and communications, channel estimation, and MIMO systems with Rydberg receivers, as the paper notes in its conclusion."],"fun_headline_variants":["Multi-chirp waveform sharpens Rydberg radar for moving targets","Two chirps beat one for atomic quantum ranging","Splitting chirps resolves Rydberg ambiguity for moving targets","Rydberg receiver gets two chirps to see moving targets clearly","Multi-chirp AFDM clears Doppler blur for Rydberg sensing"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The whole measurement model rests on the strong local-oscillator approximation |E_l| >> |E_s|, which lets the Rydberg optical response be linearized to a cosine of the phase difference; if this linearization breaks when the chirp sweeps the RF frequency across the atomic resonance, the fluctuation-frequency extraction and all subsequent delay-Doppler estimates collapse.","fun_headline_variants_meta":{"raw":{"variants":["Multi-chirp waveform sharpens Rydberg radar for moving targets","Two chirps beat one for atomic quantum ranging","Splitting chirps resolves Rydberg ambiguity for moving targets","Rydberg receiver gets two chirps to see moving targets clearly","Multi-chirp AFDM clears Doppler blur for Rydberg sensing"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000543,"raw_usage":{"total_tokens":2476,"prompt_tokens":819,"completion_tokens":1657,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":563,"completion_tokens_details":{"reasoning_tokens":1568}},"tokens_in":563,"tokens_out":1657,"duration_ms":11200,"temperature":1.0,"reasoning_tokens":1568,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-07-31T23:18:26.887841+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A controlled experiment with a single moving reflector in an anechoic chamber: transmit MC-AFDM with two known post-chirps through a Rydberg receiver, record the output voltage, and compare the two measured fluctuation frequencies against equation (32) using independently measured delay and Doppler. If the extracted frequencies deviate from the predicted linear relation by more than the noise floor, or if the least-squares delay-Doppler estimate is biased, the linearized cosine model (15) is falsified.","supporting_citations":[],"review_version":1}