{"id":"4f25606b-e637-47a3-949c-a73336296b12","arxiv_id":"2501.19142","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"An AFDM-based ISAC system with new sensing metrics tolerates Doppler shifts several times larger than OFDM at matched spectral efficiency.","lead":"This paper proposes a wireless ISAC system built on the AFDM chirp waveform, adding two new metrics for sensing efficiency and sensing outage. The system is shown to handle Doppler shifts more than five times larger than OFDM for the same spectral efficiency, which is relevant for high-mobility scenarios.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Estimation claim rests on an unvalidated early-late heuristic (Eq. 63); a wrong branch shifts Doppler by ~Δf, so the unambiguous-Doppler extension is not established.","rationale":"The strongest claim has two parts: the 5× improvement in maximum tolerable Doppler, and the estimation method that exceeds the subcarrier-spacing limit. The first part is an analytical consequence of Eqs. (35)–(37) and is plausible; the trade-off derivation and the k0 calculation are internally consistent. The second part is where the load-bearing weakness lies. The proposed estimator identifies the integral Doppler α from a peak location and the fine Doppler b from a DFT bin, but the two must be spliced through β, which can be one of two integers. The decision rule (Eq. 63) is the only mechanism for disambiguation, and it is introduced as a heuristic. Since the paper provides no derivation, no error analysis, and no simulation that specifically exercises the two-candidate case, the central claim that the unambiguous Doppler can 'break through the limitation of subcarrier spacing' is not yet established. The periodic-ambiguity delta approximation (Eq. 57) is part of the same risk: the branch decision compares two sidelobe-level cells, where the neglected µ_k sidelobes are directly relevant. This does not invalidate the paper's analytical contributions, but it does justify the reader's CONDITIONAL verdict. The proposed concrete test would settle whether the heuristic is reliable in the relevant operating regime; until then, the estimation result should be treated as conditional.","tokens_in":29533,"tokens_out":12246,"duration_ms":124058,"concrete_test":"Implement Algorithm 1 with Table II parameters and a single target whose true Doppler is chosen so that Eq. (62) has two candidate β values (e.g., α=1, b≈±0.4). For each input SNR from 0 to 20 dB, run at least 10^3 random QAM symbol frames and record whether Eq. (63) selects the correct β; also record CFAR detection success. If the early-late error rate is non-negligible (e.g., >1%) in the regime where the peak is detected, the unambiguous-Doppler claim fails. Repeat with a deterministic (constant-modulus) symbol block to isolate the effect of the delta assumption in Eq. (57).","verdict_should_be":"UNCHANGED","load_bearing_attack":"Algorithm 1's ability to measure Doppler beyond subcarrier spacing depends on correctly choosing the integer β when two candidates are possible in Eq. (62). Equation (63) selects β by comparing the amplitudes at p_i−1 and p_i+1, but this early-late rule is introduced without derivation or analysis. The two candidates correspond to fractional-Doppler values a_i with opposite signs, so the rule is a sign test on the sub-cell leakage; its reliability is not guaranteed. It further presupposes that the periodic ambiguity function µ_k is a delta (Eq. 57), whereas for finite N and random QAM symbols µ_k has O(1) sidelobes relative to an O(√N) peak. The adjacent cells p_i±1 are exactly where leakage and random-symbol sidelobes compete, so the decision can be noise-dominated even at high SNR. A wrong branch changes the Doppler estimate by Δf'=B/(N+Ncp)≈Δf·N/(N+Ncp), i.e., roughly one subcarrier spacing, destroying the claimed unambiguous-Doppler advantage. No simulation in Section VI exercises Algorithm 1 for a target whose β is genuinely two-valued; Fig. 8 uses a separate ML estimator. The central sensing-estimation claim is therefore unsupported as stated.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes an AFDM-based ISAC system, introduces two new sensing metrics (sensing spectral efficiency, SSE, and sensing outage probability, SOP), derives analytical relations between these metrics and the AFDM parameters c1, N, and Ncp, and claims that AFDM-ISAC can achieve more than five times the maximum tolerable Doppler of OFDM-ISAC at equal spectral efficiency and delay tolerance. It further proposes an estimation algorithm (Algorithm 1) that is claimed to estimate delay and the integral and fractional parts of normalized Doppler in the AFT-Doppler domain, with unambiguous Doppler extending beyond the subcarrier-spacing limit.","tokens_in":29756,"tokens_out":14389,"duration_ms":133778,"significance":"If the claims hold, the paper offers a practical waveform for high-mobility ISAC with substantially wider unambiguous velocity coverage, together with a metric framework that links sensing performance to communication-like metrics. The analytical relations in Eqs. (35)-(44) are derived in closed form without curve fitting, and the 5.6 times Doppler ratio follows algebraically from the derived fd,max expressions. The SSE/SOP metrics provide an intuitive way to discuss ISAC trade-offs, and the PSLR results under high Doppler are a concrete improvement over the OFDM-ISAC baseline. The weakest part is the parameter-estimation claim: the core early-late decision rule is unproven, and the proposed estimator is not validated in the simulation section for the regime in which the claimed unambiguous-Doppler extension matters.","major_comments":[{"comment":"The early-late criterion is introduced without derivation or analysis. When β_i has two candidate integers, a wrong choice changes the estimated Doppler by approximately Δf' = B/(N+Ncp) ≈ Δf·N/(N+Ncp), i.e., roughly one subcarrier spacing, which destroys the claimed unambiguous-Doppler extension. The criterion is a sign test on the asymmetry between |Z^l_i_F[ar{p}_i−1,\\bar{k}_i]| and |Z^l_i_F[\\bar{p}_i+1,\\bar{k}_i]|, but no proof is given that this asymmetry is monotone in a_i or robust to noise. Moreover, the adjacent cells are exactly where the approximation µ_k[p,m] ≈ c0δ[p]δ[m] (Eq. (57)) is least reliable for random QAM symbols with finite N. This is a load-bearing gap for the central estimation claim.","section":"Section V-B, Eq. (63), Algorithm 1"},{"comment":"SSE is defined as \"how much interested information on targets can be actually obtained by sensing,\" but the formula in Eq. (21) is computed conditional on the correct sub-cell being determined. Lemma 1 explicitly assumes this, and no probability of incorrect sub-cell determination enters Eq. (21); the estimation-error probability appears only in SOP (Eq. (24)). Therefore Eq. (21) is an upper bound on the actual mutual information, not the actual information, and all SSE-based comparisons (Example 3, Figs. 4 and 5(a)) overstate the sensing efficiency. The definition and the interpretation of SSE should be revised, or the text should explicitly state that SSE is an optimistic bound.","section":"Section III-B, Definition 1, Lemma 1, Eqs. (17) and (20)-(21)"},{"comment":"The proposed estimation algorithm is not validated for the scenario that motivates it. Fig. 8 reports velocity RMSE for an ML estimator, not for Algorithm 1, and uses a different frame configuration (Nc=2560, Nsym=1) than the rest of the paper (Table II: N=2560, Nsym=32). No simulation in Section VI exercises Algorithm 1 for a target whose normalized Doppler ν_i leads to two candidate values of β_i in Eq. (62), so the claimed breaking of the subcarrier-spacing limit is not empirically supported.","section":"Section VI, Figs. 6-8"}],"minor_comments":[{"comment":"The word \"genaral\" should be \"general\".","section":"Section III-A"},{"comment":"The text \"Iuput\" should be \"Input\", and \"Extrate\" should be \"Extract\".","section":"Algorithm 1"},{"comment":"The expression for k0 lacks parentheses and is ambiguous; it should read k0 = ((2ξv+2)ηcp − (2ξv+1))/(ηcp(1−ηcp)).","section":"Eq. (43)"},{"comment":"The word \"contant\" should be \"constant\".","section":"Section IV-B (2)"},{"comment":"The word \"Sweiling\" should be \"Swerling\".","section":"Remark 1"},{"comment":"The notation \"ZlF = ZlF\" is confusing; the left-hand side should be defined explicitly as the AFT-Doppler matrix.","section":"Eq. (56)"}],"recommendation":"major_revision","confidential_remarks":"The analytical trade-off results and the 5.6 times Doppler ratio are sound and constitute the strongest contribution. The main risk is the unvalidated early-late heuristic in the estimation section; the authors should either provide a rigorous derivation or extensive simulations of Algorithm 1 for the genuinely two-valued β case. The SSE definition should also be corrected to avoid overclaiming that it represents actual obtained information. The conference version [1] is acknowledged, so there is no novelty-disclosure concern."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"First thing to know: if you take away the estimation section, this is a solid waveform-level ISAC paper. The new SSE/SOP metric pair is original, the analytical relations in Section IV are correctly derived, and the 5.6x Doppler advantage over OFDM follows from those formulas rather than from fitted constants. The delay/Doppler derivations (35)-(44) check out. The paper also deserves credit for showing AFDM image SNR does not collapse at integer Doppler, where OFDM-ISAC dips badly.\n\nSoft spots, in order of importance. The central claim that the estimator can break the subcarrier-spacing limit relies on Eq. (63), an early-late rule for choosing between two candidate integral-Doppler values. As far as I can tell, it is introduced without derivation or analysis, and it is exactly the regime where finite-N ambiguity sidelobes compete with leakage: the p_i±1 cells are not noise-free. A wrong branch shifts the Doppler estimate by about one subcarrier spacing, so the advertised unambiguous-Doppler extension is not established. Moreover, Section VI never simulates Algorithm 1 for a target whose beta is genuinely two-valued; Fig. 8 uses an ML estimator, not Algorithm 1. That is a real gap between claim and evidence.\n\nThe SSE definition has a related but smaller issue. Lemma 1 conditions on the correct sub-cell being found, so the 'actually obtained' information is optimistic in exactly the noisy conditions where SOP matters. Since SSE and SOP are meant to play roles analogous to CSE and BER, splitting efficiency and reliability this way is defensible, but the phrase 'actually obtained' overstates what the metric delivers.\n\nThe finite-N approximation µ_k ≈ c0 δ[p]δ[m] is also just an approximation; for random QAM the sidelobes are O(1) against an O(√N) peak. That is probably fine for peak detection at reasonable SNR, but it is not a license to ignore adjacent-cell leakage in the early-late decision.\n\nNo circularity and no fitted constants. The cited prior work on AFDM [6], [16]-[18] is acknowledged. This paper is worth a serious referee: the metric framework and trade-off analysis are reusable, and the estimation claim is important if it can be made rigorous. I would accept it for review but push hard for a derivation or Monte-Carlo validation of Eq. (63), plus a simulation of Algorithm 1 in the ambiguous-beta regime. If that cannot be supplied, the claim should be downgraded to heuristic.","headline":"The AFDM-ISAC metric analysis is mostly sound and the 5.6x Doppler comparison is algebraically real, but the unambiguous-Doppler estimation claim rests on an unproven early-late heuristic and no simulation exercises it.","tokens_in":30291,"tokens_out":3368,"would_cite":false,"duration_ms":36341,"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":"The paper claims that an AFDM chirp waveform can more than quintuple the Doppler tolerance of an OFDM-based integrated sensing and communications system at equal spectral efficiency, and can estimate Doppler beyond the subcarrier-spacing…","keywords":["affine frequency division multiplexing","integrated sensing and communications","sensing spectral efficiency","sensing outage probability","Doppler estimation","unambiguous Doppler","high-mobility channels","AFT-Doppler domain"],"falsifier":"Run the proposed estimator on a single simulated target whose true normalized Doppler sits exactly at a candidate-integer boundary (where $\\hat{\\beta}_{\\mathrm{max}} = \\hat{\\beta}_{\\mathrm{min}}+1$ and the early and late AFT-domain samples are comparable), and record the Doppler estimate; a bias of about $\\Delta f' = B/(N+N_{\\mathrm{cp}})$ in a nontrivial fraction of trials would show the Eq. (63) branch choice fails. A second check replaces random QAM symbols with a constant-modulus or short deterministic block and observes whether the peak in $\\mathbf{Z}_l^F$ remains delta-like; if the peak spreads, the Eq. (57) approximation breaks.","tokens_in":29281,"feed_emoji":"📡","tokens_out":5773,"duration_ms":51365,"temperature":0.7,"pith_summary":"This paper argues that the affine frequency division multiplexing (AFDM) waveform, a chirp-based multicarrier modulation, can serve as an integrated sensing and communications (ISAC) waveform that outperforms OFDM in high-mobility settings. It introduces two sensing metrics—sensing spectral efficiency (SSE) and sensing outage probability (SOP)—that mirror communication spectral efficiency and bit error rate, and it derives closed-form relations between these metrics and the AFDM parameters. The central quantitative claim is that with appropriate parameters AFDM-ISAC can tolerate more than five times the maximum Doppler of OFDM-ISAC at equal spectral efficiency and equal maximum delay, and that the proposed estimator can measure Doppler beyond the conventional subcarrier-spacing limit. If the paper is right, high-mobility 6G systems could get a dual-function waveform that senses fast targets without sacrificing communication rate.","feed_headline":"AFDM chirp waveform widens Doppler coverage fivefold vs OFDM","feed_subtitle":"Integrated sensing and communications keeps spectral efficiency while tracking faster targets beyond subcarrier limits.","key_machinery":"The central object is the AFDM chirp-periodic structure and the affine Fourier transform pair. The chirp parameter $c_1$ controls the cyclic shift $\\mathrm{loc}_i = \\langle 2N c_1 l_i - \\alpha_i \\rangle_N$ in the AFT domain, which couples delay with the integral part of the normalized Doppler. The receiver decouples them by applying delay-compensation matrices $\\mathbf{L}_l = \\mathrm{diag}(e^{j2\\pi pl/N})$ and searching for peaks in the AFT-Doppler matrix $\\mathbf{Z}_l^F$. The load-bearing identity is Eq. (57): when the compensated delay matches the true delay and the Doppler matches the grid, the periodic ambiguity function of the random symbols approximates a delta and a peak appears at $(l, \\langle \\alpha_i - 2N c_1 l_i \\rangle_N, \\langle N_{\\mathrm{sym}} b_i \\rangle_{N_{\\mathrm{sym}}})$. That identity converts radar-image peak positions into delay and Doppler estimates and permits splicing integral and fractional Doppler across two different normalization grids.","core_discovery":"The paper claims that choosing the AFDM chirp parameter as $c_1 = 1/[2(N_{\\mathrm{cp}}+1)]$ makes the maximum tolerable Doppler of AFDM-ISAC about 5.6 times that of OFDM-ISAC while keeping the same cyclic prefix length, the same communication spectral efficiency, and the same maximum tolerable delay. It further claims that in the affine Fourier transform (AFT)-Doppler domain the target delay and the integral and fractional parts of the normalized Doppler separate: the AFT-domain peak position carries the combination $\\langle \\alpha_i - 2N c_1 l_i \\rangle_N$, the symbol-index DFT peak carries $\\langle N_{\\mathrm{sym}} b_i \\rangle_{N_{\\mathrm{sym}}}$, and the two are spliced through Eqs. (61)--(63) to give a Doppler estimate whose unambiguous range exceeds the subcarrier-spacing limit. Numerical results show AFDM-ISAC keeping image SNR above 40 dB for normalized Doppler up to 2, where the OFDM-ISAC image drops below 0 dB, while matching OFDM and OTFS velocity RMSE.","pith_inferences":["Because the unambiguous-Doppler gain comes from the chirp parameter $c_1$ rather than from the random data, the same decoupling idea should transfer to pilot-only or deterministic AFDM frames, provided the delta approximation for the periodic ambiguity function is replaced by a deterministic reference; this suggests the randomness of communication symbols is not essential to the sensing gain.","The early-late criterion in Eq. (63) is the weakest link: an independent derivation or a maximum-likelihood branch test across the two candidate $\\hat{\\beta}_i$ values could replace it, and simulation at the branch boundary would establish the actual failure rate.","The SSE/SOP formulation invites a unified optimization of AFDM parameters against both communication outage and sensing outage, for instance minimizing a weighted outage sum subject to an efficiency constraint, which the current paper only begins to map via trade-off curves.","Comparing AFDM-ISAC with ODDM- or OTFS-ISAC under identical pilot overhead and mobility would test whether the fivefold Doppler advantage is specific to AFDM or shared by other chirp/delay-Doppler-domain waveforms."],"forward_implications":["An AFDM-ISAC design with $c_1 = 1/[2(N_{\\mathrm{cp}}+1)]$ achieves about 5.6 times the maximum tolerable Doppler of OFDM-ISAC at the same communication spectral efficiency and same maximum delay, and therefore roughly 5.6 times the sensing spectral efficiency.","The proposed estimator measures Doppler beyond the subcarrier-spacing limit: the integral part comes from the AFT-domain peak via $\\hat{\\alpha}_i = \\langle \\bar{p}_i + 2N c_1 \\hat{l}_i + N c_1 \\rangle_N - N c_1$, the fractional part from the symbol-domain peak via Eq. (60), and the two are spliced using Eqs. (61)--(63).","Under Swerling 0, AFDM-ISAC keeps image SNR above 40 dB for normalized Doppler up to 2, while OFDM-ISAC's image SNR drops below 0 dB near integral Doppler.","The derived analytical trade-offs give a design guideline: choose $c_1$, $N_{\\mathrm{cp}}$, and $N$ satisfying Eq. (47) to meet target delay/Doppler and spectral-efficiency requirements.","Under identical time-frequency resources and an ML estimator, AFDM-ISAC matches OFDM and OTFS velocity RMSE while offering a wider unambiguous Doppler range."],"supporting_citations":[{"why":"Defines AFDM, the IDAFT/DAFT pair, the chirp-periodic prefix, and the AFT-domain input-output relation with cyclic shifts loc_i used throughout.","marker":"[6]"},{"why":"Provides the OFDM-ISAC baseline and the symbol-wise division sidelobe suppression that AFDM-ISAC is compared against.","marker":"[3]"},{"why":"Supplies the cyclic-cross-correlation OFDM baseline and the matched-filter/ambiguity-function approach the AFT-Doppler estimator extends.","marker":"[4]"},{"why":"Gives the time-domain received-signal model with delay and Doppler that the sensing channel in Eq. (48) is built on.","marker":"[29]"},{"why":"Supplies radar fundamentals: delay-Doppler resolution cells, CRB, Swerling models, and detection/estimation metrics used to define SSE and SOP.","marker":"[20]"},{"why":"Provides the information-theoretic detection mutual information formula that SSE adapts to a given ISAC waveform.","marker":"[28]"},{"why":"Supplies the OTFS-ISAC baseline and ML velocity estimator used for the RMSE comparison.","marker":"[34]"}],"fun_headline_variants":["AFDM-ISAC widens Doppler coverage 5.6x over OFDM","Integrated sensing and comms using AFDM extends Doppler range","Chirp-based ISAC breaks Doppler limit, boosts sensing range","AFDM waveform boosts unambiguous Doppler in ISAC systems","Beyond OFDM: AFDM-ISAC widens Doppler coverage for fast targets"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The whole unambiguous-Doppler extension rests on the early-late rule in Eq. (63) choosing the correct candidate Doppler integer when two are possible, and on the periodic ambiguity function of the random symbol block being close to a delta; if the rule or the delta approximation fails, the estimated Doppler can jump by roughly one subcarrier spacing of the AFT-Doppler grid.","fun_headline_variants_meta":{"raw":{"variants":["AFDM-ISAC widens Doppler coverage 5.6x over OFDM","Integrated sensing and comms using AFDM extends Doppler range","Chirp-based ISAC breaks Doppler limit, boosts sensing range","AFDM waveform boosts unambiguous Doppler in ISAC systems","Beyond OFDM: AFDM-ISAC widens Doppler coverage for fast targets"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000645,"raw_usage":{"total_tokens":2963,"prompt_tokens":946,"completion_tokens":2017,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":562,"completion_tokens_details":{"reasoning_tokens":1924}},"tokens_in":562,"tokens_out":2017,"duration_ms":15398,"temperature":1.0,"reasoning_tokens":1924,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-09T21:10:10.935358+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the proposed estimator on a single simulated target whose true normalized Doppler sits exactly at a candidate-integer boundary (where $\\hat{\\beta}_{\\mathrm{max}} = \\hat{\\beta}_{\\mathrm{min}}+1$ and the early and late AFT-domain samples are comparable), and record the Doppler estimate; a bias of about $\\Delta f' = B/(N+N_{\\mathrm{cp}})$ in a nontrivial fraction of trials would show the Eq. (63) branch choice fails. A second check replaces random QAM symbols with a constant-modulus or short deterministic block and observes whether the peak in $\\mathbf{Z}_l^F$ remains delta-like; if the peak spreads, the Eq. (57) approximation breaks.","supporting_citations":[{"cited_title":"Affine frequenc y division multiplexing for next generation wireless communications ,","cited_arxiv_id":null,"evidence_quote":"Defines AFDM, the IDAFT/DAFT pair, the chirp-periodic prefix, and the AFT-domain input-output relation with cyclic shifts loc_i used throughout."},{"cited_title":"Waveform design and signal pro cessing aspects for fusion of wireless communications and radar sen sing,","cited_arxiv_id":null,"evidence_quote":"Provides the OFDM-ISAC baseline and the symbol-wise division sidelobe suppression that AFDM-ISAC is compared against."},{"cited_title":"Joint radar-communication wi th cyclic prefixed single carrier waveforms,","cited_arxiv_id":null,"evidence_quote":"Supplies the cyclic-cross-correlation OFDM baseline and the matched-filter/ambiguity-function approach the AFT-Doppler estimator extends."},{"cited_title":"Integrating l ow- complexity and flexible sensing into communication systems ,","cited_arxiv_id":null,"evidence_quote":"Gives the time-domain received-signal model with delay and Doppler that the sensing channel in Eq. (48) is built on."},{"cited_title":"A Richards, J","cited_arxiv_id":null,"evidence_quote":"Supplies radar fundamentals: delay-Doppler resolution cells, CRB, Swerling models, and detection/estimation metrics used to define SSE and SOP."},{"cited_title":"Information-theoretical approach to integrated pulse-Doppler radar and communication systems,","cited_arxiv_id":null,"evidence_quote":"Provides the information-theoretic detection mutual information formula that SSE adapts to a given ISAC waveform."},{"cited_title":"On the effective- ness of OTFS for joint radar parameter estimation and commun ication,","cited_arxiv_id":null,"evidence_quote":"Supplies the OTFS-ISAC baseline and ML velocity estimator used for the RMSE comparison."}],"review_version":1}