{"id":"85a1f89e-5342-46b4-aeb6-a22fc3e9c3c8","arxiv_id":"2507.03537","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"The paper derives AFDM transmission and detection for wideband doubly-dispersive channels with time-scaling effects, and shows optimized chirp parameters improve BER over existing schemes.","lead":"A transmission scheme for affine frequency division multiplexing is extended to wideband channels where Doppler shifts change with frequency, and new chirp parameters plus a low-complexity detector are derived. The paper shows, by simulation, that the tuned AFDM scheme outperforms OFDM, OCDM, OTFS and standard AFDM in such channels.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Eq. (20) has an unresolved m-dependent ψ term, and the POSP support bounds (30)-(32) are unproved; because the optimized c1 in (45) is designed from those bounds, the central sparsity/diversity claim is not yet established.","rationale":"The reader's conditional verdict already targets the POSP approximation and the questionable ψ term in Eq. (20); my review identifies the same load-bearing weakness and sharpens it: the ψ term is not merely imported without proof, but is left with an undefined m index, and its coefficient appears inconsistent with a direct expansion of (7)-(13). Because the optimized c1 in (45) is derived from support intervals that depend on this term, the central claim inherits the uncertainty. I do not move the verdict to REJECT because no calculation in the paper or in my review yet demonstrates an actual overlap failure, and the paper has some independent support, notably the Fig. 2 POSP-vs-exact comparison and the PEP/ML matching in Fig. 3. However, those checks use the same channel-formula chain, so they do not by themselves validate the exact support structure. A direct numerical test from the time-domain model would settle the matter; until then, the appropriate disposition remains CONDITIONAL, matching the reader's verdict.","tokens_in":24120,"tokens_out":12109,"duration_ms":148452,"concrete_test":"Directly compute the exact DAF-domain matrix in (19) by substituting x=e_q into the sampled time-domain model (7)-(13), fixing the undefined m in ψ_{m,t} to the column index q. Use a small two-path case: N=64, ℓ_1=0, ℓ_2=1, α_1=-α_max, α_2=+α_max with α_max=10^-4, and set c1 by (45). For every row p, record whether the peak-normalized supports of |H_1(p,:)| and |H_2(p,:)| overlap above a threshold of 10^-3. If any overlap occurs, the bounds (30)-(32) are not the exact supports used to derive (45), so path separation and the resulting diversity claim are not established. Repeat with the ψ coefficient in (20)/(24c) replaced by (1+α_i)N to separate transcription error from POSP error.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim that the optimized chirp parameter (45) yields a sparse DAF-domain channel and diversity gain rests on the POSP support intervals in Section II-D. That step is load-bearing and currently insecure. First, Eq. (20) defines F_i(p,q) as the (p,q)-th channel coefficient but contains ψ_{m,t'_i} with no defined m; the expression is not a well-defined function of (p,q), and the stationary-phase calculation in (24)-(26) then carries the same ambiguous m-dependence into the support bounds. Second, a direct expansion of the continuous basis (7)-(8) through the sampled time-domain model (13) gives a ψ coefficient proportional to (1+α_i)N, not α_iN as written in (20) and (24c); since α_i is at most 10^-4, the two versions differ by orders of magnitude in the widening term used to obtain (32). If either issue is real, intervals (30)-(32) do not describe the exact channel, so (42)-(45) is not guaranteed to separate paths, and the PEP/diversity and BER conclusions in Figs. 3, 4, 10, 11, and 13 do not follow from the claimed mechanism.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes an affine frequency division multiplexing (AFDM) transmission scheme for wideband doubly-dispersive channels with time-scaling Doppler effects. It introduces a chirp-periodic prefix/suffix (CPP/CPS) frame structure, derives a discrete affine Fourier (DAF)-domain input-output relation, uses the principle of stationary phase (POSP) to claim sparsity of the DAF-domain channel, optimizes the AFDM chirp parameter c1, derives a pairwise error probability (PEP) bound, and proposes a cross-domain distributed orthogonal approximate message passing (CD-D-OAMP) detector with a state-evolution analysis. The claims are supported by simulations for underwater acoustic and terahertz channels, comparing the proposed AFDM with OFDM, OCDM, OTFS, and narrowband-AFDM.","tokens_in":24392,"tokens_out":7399,"duration_ms":90589,"significance":"The problem is well motivated: time-scaling effects are known to break the standard narrowband Doppler model, and most existing AFDM/OTFS analyses ignore them. If the derivation is made rigorous, the paper would be a solid contribution: it provides a concrete CPP/CPS frame structure, a closed-form chirp parameter design, a low-complexity distributed detector with a useful complexity table, and broad simulation evidence. The PEP and state-evolution analyses are useful complements. However, the central technical claim—that the optimized chirp parameter in Eq. (45) separates paths and yields sparsity and diversity—depends on the POSP-based support intervals in Eqs. (30)-(32), and those are not yet rigorously established. The significance is therefore conditional on fixing the derivation.","major_comments":[{"comment":"The derivation of the DAF-domain channel is not reproducible as written. The piecewise function ψ_{m,t} in Eq. (9) is defined through breakpoints indexed by ρ = 1, ..., \\tilde C with \\tilde C = 2Nc1 in Eq. (10), but the optimized c1 in Eq. (45) is not guaranteed to make 2Nc1 an integer. In Eq. (20), F_i(p,q) contains ψ_{m,t'_i} with no definition of m and no summation over m, so F_i(p,q) is not a well-defined function of (p,q). The same undefined symbol enters θ_{p,q}(n) in Eq. (24c), the stationary point in Eq. (26), and the support bounds in Eqs. (31)-(32). The authors should define the sampled ψ term explicitly, including the correct summation index and the correct coefficient, and should either restrict c1 to values for which 2Nc1 is integer or re-derive the piecewise construction for non-integer c1.","section":"II-B/II-C, Eqs. (9), (10), (20), (24c)"},{"comment":"The POSP approximation is imported from continuous-time radar analysis and applied to the discrete quadratic-phase sum in Eq. (23) without a proof that it is valid for this discrete, finite-length sum. The approximations in Eqs. (27)-(28) and the resulting support intervals in Eqs. (29)-(32) are load-bearing for the chirp parameter design in Section III, but no validity condition on K = 2c1(α_i^2 + 2α_i), N, c1, or α_i is provided. The single illustration in Fig. 2 for one row and one channel realization is not sufficient. A derivation, or at least a systematic numerical verification over the parameter ranges used in Figs. 3, 4, 10, 11, and 13, should be supplied.","section":"II-C/II-D, Eqs. (23)-(34)"},{"comment":"The transition from the general no-overlap condition (41a) to the closed-form c1 in Eq. (45) relies on the additional assumption min(ℓ_j − ℓ_i) = 1, which is described as the dense-delay case. This assumption is not derived from the channel model and is not stated as an explicit system condition in the simulation setup. If the actual minimum delay separation differs from one sample, the formula may not give the intended separation guarantee; and if the POSP intervals are inaccurate, the no-overlap condition in Eq. (40) may not be sufficient at all. The paper should state this assumption explicitly and verify that the simulated channels satisfy it.","section":"III, Eqs. (42)-(45)"},{"comment":"The PEP bound in Eq. (61) is only meaningful if the rank R of Ω_{x,\\hat x} is known; in particular, the diversity claim requires R to equal the number of resolvable paths P (or at least to be established). The paper does not prove that the optimized c1 in Eq. (45) achieves R = P for the time-scaled wideband channel. Without this rank analysis, the diversity conclusion drawn from Figs. 3 and 4 is not established by the PEP argument. Please add a rank analysis or state clearly the conditions under which Eq. (61) holds.","section":"IV, Eqs. (60)-(61)"}],"minor_comments":[{"comment":"The Chernoff bound should read Q(x) ≤ exp(−x^2/2); as printed, the exponent '1/2x2' has the wrong sign.","section":"IV, Eq. (60)"},{"comment":"The phrase 'in the literatures' should be 'in the literature'.","section":"Abstract"},{"comment":"The significant-region width Nv is introduced without a quantitative selection criterion, and it directly affects the optimized c1 in Eq. (45). Please state how Nv is chosen in the simulations.","section":"II-D, Eq. (33)"},{"comment":"The symbol |\\tilde D_c| in the D-OAMP complexity row should be defined in the table caption or immediately before the table, since the definition currently appears only in the body text.","section":"Table I"},{"comment":"The state-evolution analysis uses a Monte Carlo approximation for the nonlinear function f_D; this is acknowledged in the text, but the limitation should also be stated in the conclusion where the state-evolution result is summarized.","section":"V-B, Eq. (83)"}],"recommendation":"major_revision","confidential_remarks":"The mathematical issue is central but appears fixable. The undefined m in Eq. (20) and the unproved POSP support intervals are not merely presentation problems: the optimized chirp parameter, the no-overlap condition, the PEP bound, and the diversity conclusions all depend on them. I would ask the authors to provide a rigorous derivation or a validated bound for the support intervals, and to check whether the proposed c1 is consistent with an integer-valued 2Nc1. If those points can be resolved, the paper could be acceptable for publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things you should know about arXiv:2507.03537. First, it is the first paper I know that takes AFDM into wideband doubly-dispersive channels with time-scaling effects, and it comes with a concrete chirp-parameter design and a distributed detector. Second, the sparsity analysis that supports the whole design has a specific algebraic error you can verify in ten minutes.\n\nWhat is genuinely new: the CPP/CPS frame structure is a sensible adaptation of chirp-periodic AFDM to time-scaled channels, and the CD-D-OAMP detector is a reasonable extension of cross-domain OAMP with parallel-computing advantages. The state-evolution verification is honest—the authors admit the Monte Carlo approximation of f_D. The BER comparisons against OFDM, OCDM, OTFS, and narrowband AFDM are thorough, and the complexity analysis is useful.\n\nThe soft spot is Eq. (20). The exponent contains ψ_{m,t'_i} with an m that does not appear as a summation index in a channel-matrix entry; presumably m should be q, but as written the expression is not a well-defined function of (p,q). More seriously, carrying the continuous chirp-periodic basis (7)-(8) through the time-scaled sampling (13) gives a ψ coefficient proportional to (1+α_i)N, not α_i N. With α_i ≤ 10^-4, that is a factor of 10^4. The stationary-phase support intervals (30)-(32) and the strengthened bounds (32) inherit this error, so the interval width L_i in (34) is off by roughly 1/α_i. The chirp-parameter design in (42)-(45) is built on those intervals. If the correct coefficient is (1+α_i)N, the DAF channel is not sparse in the claimed sense, and the PEP/diversity argument in Section IV does not follow from the stated mechanism. The paper's own note that POSP derivations are omitted 'due to space limitation' is a warning sign when that step is load-bearing. Fig. 2 shows a good match in one example, but that does not fix an equation error.\n\nThe detector contribution may survive a correction, and the wideband-AFDM idea is worth exploring. I would not desk-reject this. Send it to referees with a specific request to check Eq. (20) and the support bounds; the authors need to redo the c1 derivation. The paper is aimed at the AFDM/OTFS community and people working on wideband underwater and THz links. A careful reader will get a useful detector and a good example of why stationary-phase approximations need to be checked against the discrete model. I would not cite it until the sparsity analysis is corrected.","headline":"Wideband AFDM paper with a useful detector and a serious algebra error in the sparsity analysis; the c1 design is built on the wrong ψ coefficient.","tokens_in":24926,"tokens_out":9672,"would_cite":false,"duration_ms":99816,"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":"One optimized chirp parameter keeps the AFDM channel sparse under wideband time-scaling Doppler and outperforms four rival modulations in simulation.","keywords":["affine frequency division multiplexing","wideband doubly-dispersive channels","time-scaling effects","Doppler squint","chirp parameter optimization","cross-domain distributed OAMP","sparse channel equalization","OTFS comparison"],"falsifier":"Compute the exact DAF-domain channel matrix for a small block, say $N=32$, with two paths whose delays differ by one sample and $\\alpha_{\\max}=10^{-4}$, and check whether the support intervals predicted by (30)-(32) contain the actually nonzero entries; if the paths overlap in the exact matrix under the $c_1$ of (45), the no-overlap premise fails.","tokens_in":23939,"feed_emoji":"📡","tokens_out":6235,"duration_ms":68202,"temperature":0.7,"pith_summary":"This paper tries to establish that affine frequency division multiplexing can be made to work in wideband doubly-dispersive channels where the Doppler effect scales time rather than just shifting frequency. The authors add a chirp-periodic prefix and suffix so AFDM symbols stay chirp-periodic under time scaling, derive the input-output relation in the discrete affine Fourier domain, and choose the chirp parameter $c_1$ so that different propagation paths occupy disjoint bins. They also design a cross-domain distributed OAMP detector and analyze its state evolution. If the claims hold, AFDM with the optimized chirp parameter would beat OFDM, OCDM, OTFS, and narrowband AFDM in such channels at the same SNR.","feed_headline":"A chirp-rate tweak keeps AFDM's channel sparse under wideband Doppler","feed_subtitle":"New chirp-rate rule and prefix design keep paths separated, beating four rivals in wideband simulations.","key_machinery":"The load-bearing object is the AFDM chirp parameter $c_1$, which controls the quadratic phase in both the inverse DAF transform and the received channel matrix. In this wideband model the Doppler scale $\\alpha_i$ enters the channel as a time-dependent delay, so $c_1$ must absorb a term proportional to $\\alpha_i n^2$; the optimized value in (45) is chosen so that the POSP-derived support intervals of distinct paths do not overlap. The CPP and CPS prefixes keep the received AFDM symbol chirp-periodic inside the observation window, while the CD-D-OAMP detector alternates a distributed LMMSE estimate in the sparse time domain with symbol-by-symbol detection in the DAF domain, using the unitary DAF transform to cross between them.","core_discovery":"The central claim is that in time-scaled wideband doubly-dispersive channels, an AFDM system equipped with the CPP/CPS frame and the chirp parameter $c_1$ of (45) keeps a sparse discrete affine Fourier domain channel and thereby retains full path diversity. With narrowband parameters, Doppler scaling smears each path across many delay-Doppler bins; the optimized $c_1$ makes the per-path support intervals $[Q_{\\ell_i,\\alpha_i},\\tilde{Q}_{\\ell_i,\\alpha_i}]$ disjoint for resolvable paths. The paper supports this by a stationary-phase approximation of the channel coefficients, a pairwise-error-probability bound whose slope matches maximum-likelihood simulations, and BER comparisons showing the proposed system below OFDM, OCDM, OTFS, and narrowband AFDM in underwater and THz wideband settings.","pith_inferences":["Beyond the paper: if the POSP support intervals remain accurate for finite $N$, the same single-parameter design recipe could be applied to other chirp-based waveforms, by picking $c_1$ so that the $\\alpha_i$-dependent support widths are disjoint.","Beyond the paper: the CD-D-OAMP idea is not tied to AFDM's specific transform; any unitary cross-domain pair with one sparse domain could use the same distributed LMMSE plus symbol-wise projection loop, so the detector recipe may transfer to other wideband waveforms.","Beyond the paper: a stress test the paper does not run is to push $\\alpha_{\\max}$ toward $1/(4N)$, where (46) binds, and check whether the BER advantage over OTFS disappears at the predicted block size.","Beyond the paper: the paper treats $\\alpha_{\\max}$ as small (at most $10^{-4}$ in its examples); at larger Doppler scale factors the support width $L_i$ grows linearly in $N$, so the authors' own formula implies that either $N$ must shrink or the sparsity-based receiver will lose its edge."],"forward_implications":["With the optimized $c_1$ and CPP/CPS framing, wideband Doppler scaling no longer destroys DAF-domain sparsity: each resolvable path lands in its own support interval, so path diversity can be collected.","The PEP bound in (61)-(62) gives a full-diversity slope in SNR for the proposed parameters; the paper verifies it with ML detection for $P=2,3,4$ paths.","The CD-D-OAMP detector with $C$ groups reduces per-iteration complexity from $O(N^3)$ to $O(CN_c^3 + CN_c^2|D_c| + N\\log N + NQ)$, with only a small BER penalty as $C$ grows.","In simulations, AFDM with the wideband-optimized $c_1$ achieves lower BER than OFDM, OCDM, OTFS, and narrowband AFDM under both underwater acoustic ($\\alpha_{\\max}=10^{-4}$) and THz wireless ($\\alpha_{\\max}=4.6\\cdot10^{-7}$) channels.","The design constraint (51) tells the system designer the usable range of $N$ for given $\\alpha_{\\max}$ and delay spread; beyond it, the path supports become too wide to separate."],"supporting_citations":[{"why":"Supplies the wideband channel model with time-dependent delay $\\tau_i(t)=\\tau_i-\\alpha_i t$ and non-uniform Doppler shifts that the whole analysis starts from.","marker":"[1]"},{"why":"Defines AFDM, the DAF transform, the chirp-parameter conditions in the narrowband case, and the input-output relationship that this paper extends to wideband.","marker":"[9]"},{"why":"Provides the OFDM baseline that suffers from Doppler spread in wideband channels; the paper compares its BER against the proposed AFDM.","marker":"[17]"},{"why":"Provides the OTFS baseline with Doppler squint effect that destroys delay-Doppler sparsity; the paper compares against it in wideband settings.","marker":"[19]"},{"why":"Provides the OCDM baseline that cannot adjust its chirp rate for wideband channels; used in the BER comparisons.","marker":"[22]"},{"why":"Supplies the OAMP detector whose LMMSE module and convergence properties the proposed CD-D-OAMP builds on and compares with.","marker":"[32]"},{"why":"Supplies the cross-domain iterative detection principle and the unitary-transform argument used to justify the CD-D-OAMP convergence.","marker":"[39]"},{"why":"Is the source of the stationary-phase approximation used to derive the sparse support intervals and the chirp-parameter optimization.","marker":"[46]"}],"fun_headline_variants":["Chirp-rate tweak keeps AFDM sparse in time-scaled wideband","AFDM's sparse channel survives Doppler scaling via new chirp rule","New chirp parameter preserves AFDM diversity under wideband time-scaling","Beating OFDM and OTFS: AFDM chirp tuning handles time-scaling","Time-scaling won't smear AFDM paths with optimized chirp and prefix"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The whole parameter design assumes that the stationary-phase approximation, imported without derivation, correctly predicts which delay-Doppler bins each path occupies; if that approximation is inaccurate for finite block sizes, the optimized chirp parameter will not separate paths and the claimed diversity gain disappears.","fun_headline_variants_meta":{"raw":{"variants":["Chirp-rate tweak keeps AFDM sparse in time-scaled wideband","AFDM's sparse channel survives Doppler scaling via new chirp rule","New chirp parameter preserves AFDM diversity under wideband time-scaling","Beating OFDM and OTFS: AFDM chirp tuning handles time-scaling","Time-scaling won't smear AFDM paths with optimized chirp and prefix"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000372,"raw_usage":{"total_tokens":2019,"prompt_tokens":1008,"completion_tokens":1011,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":624,"completion_tokens_details":{"reasoning_tokens":909}},"tokens_in":624,"tokens_out":1011,"duration_ms":12600,"temperature":1.0,"reasoning_tokens":909,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T20:07:40.298682+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the exact DAF-domain channel matrix for a small block, say $N=32$, with two paths whose delays differ by one sample and $\\alpha_{\\max}=10^{-4}$, and check whether the support intervals predicted by (30)-(32) contain the actually nonzero entries; if the paths overlap in the exact matrix under the $c_1$ of (45), the no-overlap premise fails.","supporting_citations":[{"cited_title":"Multicarrier communication over underwater acoustic channels with nonu niform Doppler shifts,","cited_arxiv_id":null,"evidence_quote":"Supplies the wideband channel model with time-dependent delay $\\tau_i(t)=\\tau_i-\\alpha_i t$ and non-uniform Doppler shifts that the whole analysis starts from."},{"cited_title":"Effect of Doppler spr ead in OFDM- based UWB systems,","cited_arxiv_id":null,"evidence_quote":"Provides the OFDM baseline that suffers from Doppler spread in wideband channels; the paper compares its BER against the proposed AFDM."},{"cited_title":"On the Doppler squi nt effect in OTFS systems over doubly-dispersive channels: Mo deling and evaluation,","cited_arxiv_id":null,"evidence_quote":"Provides the OTFS baseline with Doppler squint effect that destroys delay-Doppler sparsity; the paper compares against it in wideband settings."},{"cited_title":"Underwater acousti c commu- nications based on OCDM for Internet of Underwater Things,","cited_arxiv_id":null,"evidence_quote":"Provides the OCDM baseline that cannot adjust its chirp rate for wideband channels; used in the BER comparisons."},{"cited_title":"Orthogonal AMP ,","cited_arxiv_id":null,"evidence_quote":"Supplies the OAMP detector whose LMMSE module and convergence properties the proposed CD-D-OAMP builds on and compares with."},{"cited_title":"Cross domain iterati ve detection for orthogonal time frequency space modulation,","cited_arxiv_id":null,"evidence_quote":"Supplies the cross-domain iterative detection principle and the unitary-transform argument used to justify the CD-D-OAMP convergence."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Is the source of the stationary-phase approximation used to derive the sparse support intervals and the chirp-parameter optimization."}],"review_version":1}