{"id":"d0d52329-a471-46dc-85cd-2fe076ea87a3","arxiv_id":"2411.09938","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"Two new AFDM schemes that add cyclic delay diversity and index modulation achieve full diversity under derived conditions and beat AFDM, CDD-AFDM, and OTFS benchmarks with a new low-complexity detector.","lead":"This paper designs two new ways to combine a chirp-based radio waveform with on-off subcarrier signaling and multiple transmit antennas, and it works out the conditions needed to get full diversity and a fast detector. If correct, the schemes give more reliable communication for high-speed trains, vehicles, and other fast-moving radio links with less complexity than full maximum-likelihood detection.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Fractional-Doppler full-diversity proof relies on a truncated DAF-domain channel (Eq. 12) without specifying k_alpha; the exact Dirichlet kernel has full support, so conditions (21)/(22) may not characterize the real system.","rationale":"The reader's verdict is CONDITIONAL, and the weakest assumption identified is exactly the fractional-Doppler band truncation after Eq. (12). My stress test agrees that this is the most load-bearing point: the full transmit diversity claim for fractional Doppler is justified only on a truncated channel model, and no k_alpha is supplied for the simulations that are supposed to confirm the claim. The concern is not that the proposed schemes necessarily fail; the simulations may still show good performance, and the exact unitary nature of each H_e,l might even salvage the rank. But the proof as written does not connect the exact channel to condition (22), and the missing k_alpha makes the fractional-Doppler results unreproducible. I do not see a separate concern that would move the verdict to ACCEPT or REJECT; the most appropriate action is to keep the CONDITIONAL verdict and require either a specification of k_alpha with a justification, or a rank/PEP check on the exact fractional-Doppler channel.","tokens_in":24317,"tokens_out":15081,"duration_ms":180283,"concrete_test":"Recompute the PEP diversity without truncation: for the parameters of Fig. 8(b) (e.g., N=64, Nt=4 or 5, P=3, Jakes-distributed fractional Doppler), form the exact N x (P*Nt) matrix Upsilon_xi-xj using F_e,l from Eq. (8) for several symbol pairs and compute its rank. If the minimum rank is below P*Nt, condition (22) is not sufficient; if it equals P*Nt, the full-diversity conclusion may survive but the truncation analysis is still not the right justification. In either case, rerun the fractional-Doppler simulation with k_alpha = 0, 1, 2 and record the high-SNR BER slope; a slope that depends on k_alpha would confirm that the unstated truncation parameter is load-bearing.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The load-bearing step is the reduction of the fractional-Doppler channel to a finite band. In Eq. (12), H_e,l is set to zero except when v = [bar v + loc_fra,e,l]_N with loc_fra inside a 2k_alpha+1 window, so the non-overlap condition (22) is proven only for a truncated matrix. The exact F_e,l in Eq. (8) is a Dirichlet kernel whose magnitude decays but is nonzero for every v; as a matrix, H_e,l is unitary up to phase factors, so its support is not a finite band. Therefore any statement about 'non-zero elements not overlapping' is not a property of the true fractional-Doppler channel. The paper never supplies k_alpha for the fractional-Doppler simulations in Figs. 8(b) and 10(b), and the rank of the true difference matrix Upsilon_xi-xj could be lower than the truncated-model rank when sidelobe overlaps make the P*Nt columns dependent. Since the full-diversity claim and the diversity order read from the BER upper bound both depend on this condition, the central claim is not established for the actual fractional-Doppler channel.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes two index-modulation-assisted AFDM schemes with multiple transmit antennas, CDD-AFDM-IM-I and CDD-AFDM-IM-II, using cyclic delay diversity to obtain transmit diversity over linear time-varying channels. It analyzes parameter settings for full diversity in both integer- and fractional-Doppler cases, derives closed-form BER upper bounds under ML detection, and develops a double-layer message-passing (DLMP) detector for large-dimensional signal detection. Simulation results compare the proposed schemes and detector against AFDM, OTFS, CDD-AFDM, MP, and MMSE benchmarks in terms of BER and computational complexity.","tokens_in":24485,"tokens_out":9352,"duration_ms":99305,"significance":"If the theoretical claims are correct, the paper provides useful design rules for extending AFDM to multi-antenna index-modulation transmission and a practical low-complexity detector. The integer-Doppler disjoint-support argument is clear, the DLMP detector is evaluated against standard benchmarks with explicit FLOP counts, and the simulated BER curves in Figs. 4 and 5 confirm the predicted diversity slopes. However, the fractional-Doppler full-diversity analysis relies on an unquantified truncation of the DAF-domain channel, and the IM-diversity example omits same-activation error events. These gaps affect the central claims about full diversity and the diversity protection provided by index modulation.","major_comments":[{"comment":"The fractional-Doppler full-diversity analysis replaces the exact DAF-domain response F_{e,l}[bar v, v] in Eq. (8), a Dirichlet kernel with nonzero values for every v, by a truncated response supported only on a band of width 2k_alpha+1 around each nominal location. Conditions (16), (18), (20), and (22) are therefore conditions on a truncated channel matrix, not on the true channel matrix H_{e,l}. The paper does not specify k_alpha for the fractional-Doppler simulations in Figs. 8(b) and 10(b), nor does it bound the out-of-band tail. Because the rank of the true difference matrix Upsilon_{xi-xj} can be smaller than that of its truncated version when sidelobe overlaps make the P*Nt columns dependent, the full-diversity claim for the fractional-Doppler channel modeled by Eq. (4) is not established. Please provide a rigorous tail bound showing that the truncation cannot change the rank, or specify k_alpha and prove that the diversity order is preserved for all channel realizations in the chosen parameter regime.","section":"Section III, Eq. (12)"},{"comment":"The high-SNR approximation in Eq. (30) omits the factor (PNt)^d that follows from Eq. (29). With q1=1/(4N0) and q2=1/(3N0), each product in Eq. (29) tends to (PNt)^d (4N0)^d / prod(kappa_i^2) and (PNt)^d (3N0)^d / prod(kappa_i^2), respectively. The displayed result should contain (PNt)^d SNR^{-d}(4^d/12 + 3^d/4)/prod(kappa_i^2). The diversity order d is unchanged, but the claimed asymptotically tight upper bound in Eq. (31) has the wrong multiplicative constant, so the theoretical curves in Figs. 4 and 5 do not correspond to the stated bound.","section":"Section IV, Eq. (30)"},{"comment":"The proof that index modulation can provide additional diversity when the full-diversity condition in (21) fails considers only pairs with different IM activation states. For a pair with the same activation state but different modulated symbols, the difference vector xi-xj contains only m nonzero entries, so the rank of Upsilon_{xi-xj} is at most m; in the N=4, m=1 example this rank is 1. Since the diversity order in Lemma 1 is the minimum over all pairs xi != xj, these same-activation symbol-error pairs determine the overall diversity order of the scheme. The example therefore does not show that CDD-AFDM-IM-I has a higher diversity order than CDD-AFDM when the full-diversity condition is violated; it shows only that index bits enjoy a larger rank when the index decision is wrong. The diversity-order claim for the whole scheme needs to be re-evaluated over all pairwise error events, and the comparison in Fig. 6 should be interpreted accordingly.","section":"Section IV-A, Eqs. (32)-(35)"}],"minor_comments":[{"comment":"The sentence 'where each group in xj and xj carriers the same index bits' should read 'where each group in xi and xj carries the same index bits.'","section":"Section IV-A, after Eq. (37)"},{"comment":"The phrase 'prefect CSIs' should be 'perfect CSI.'","section":"Section VI"},{"comment":"The acronym 'CDD-AFM-IM' appears in the contribution list; it should be 'CDD-AFDM-IM' for consistency.","section":"Section I, contributions"},{"comment":"The message from the constraint node to the indicator node is denoted both 'uc' and 'u_c^{niter}' in the surrounding text; please unify the notation.","section":"Section V, Eq. (49)"},{"comment":"The caption contains garbled text '´eff H xy'; it should display the relation y = H_eff x.","section":"Fig. 2 caption"}],"recommendation":"major_revision","confidential_remarks":"The main concerns are technical and centered on the fractional-Doppler rank argument and the incomplete diversity-order proof for IM. The self-citation [30] is used as a baseline rather than as a source of the main claims, which is appropriate. The paper fits the scope of the journal and the simulation work is substantial, but the theoretical gaps should be resolved before publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nRead this if you work on AFDM or delay-Doppler waveforms. The genuinely new pieces are the two CDD-AFDM-IM schemes, the integer-Doppler full-diversity conditions, and the DLMP detector. The paper deserves a serious referee, but not in its current form: the fractional-Doppler full-diversity proof has a hole, and the PEP bound has a missing factor.\n\nWhat is good. The two schemes are new combinations — [30] did AFDM-IM without transmit diversity, [33] did cyclic delay-Doppler shift for AFDM without IM and without analysis. The paper works out the cyclic-delay interval needed to separate paths in the DAF domain, gives closed-form BER bounds, and the simulated BER slopes in Figs. 4–5 match the claimed diversity orders. The DLMP detector is a sensible extension of message passing to jointly handle activation states and constellation symbols, and the complexity comparison against MP and MMSE is honest. The integer-Doppler case looks internally consistent.\n\nThe soft spots.\n\n1. Fractional Doppler. Eq. (12) truncates the channel matrix to a band of width 2k_alpha+1 around each nominal tap. But the exact F in Eq. (8) is a Dirichlet kernel with nonzero entries everywhere; the matrix is not banded. The non-overlap condition (22) is proved only for the truncated matrix. The true condition for full diversity is rank of the difference matrix, not non-overlap of support. The paper never supplies k_alpha for the fractional-Doppler simulations, so the full-diversity claim in Figs. 8(b) and 10(b) is not tied to a stated parameter. This is a genuine gap. It may be fixable by a proper rank argument, but as written the fractional-Doppler full-diversity result is not established.\n\n2. The PEP derivation. Eq. (29) has factors 1/(1+q kappa_i^2/(NtP)). At high SNR each factor behaves like NtP/(q kappa_i^2), so the product carries a (PNt)^d factor. Eq. (30) drops it. The diversity order d is unaffected, but the constant is wrong, so the label \"upper bound\" is not as tight as claimed.\n\n3. The claim that IM gives extra diversity when the full-diversity condition fails is demonstrated only for one small example (Nt=2, N=4, P=1). The example is suggestive, not a general proof. The abstract says \"prove\"; that is too strong.\n\nNo code or data is provided, so exact reproduction means reimplementation. That is not fatal for a theory-and-simulation paper, but it does limit how quickly the fractional-Doppler result can be checked.\n\nBottom line. The integer-Doppler CDD-AFDM-IM design and the DLMP detector are solid and useful; the fractional-Doppler analysis needs more work. I would send this to peer review, but I would ask the authors to fix Eq. (30), supply k_alpha, and either prove the fractional-Doppler rank condition without truncation or soften the claims. For a researcher in this subfield, it is worth reading and citing for the integer-Doppler results.","headline":"Real new AFDM-IM + CDD schemes and a practical detector, but the fractional-Doppler full-diversity proof is not established and the PEP bound has a missing factor.","tokens_in":25162,"tokens_out":5375,"would_cite":true,"duration_ms":54872,"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":"The paper claims that combining cyclic delay diversity with index modulation in affine frequency division multiplexing (AFDM) can achieve full transmit diversity in high-mobility channels, provided the normalized delay and Doppler paths…","keywords":["affine frequency division multiplexing","index modulation","cyclic delay diversity","transmit diversity","linear time-varying channel","double-layer message passing","bit error rate upper bound","high-mobility communications"],"falsifier":"Take a channel whose fractional Doppler shifts produce leakage outside the assumed $2k_\\alpha+1$ band (for example, two paths whose band centers satisfy (22) but whose true sinc-like tails overlap), simulate ML detection without truncating $\\mathbf{H}_{\\epsilon,\\ell}$, and check whether the high-SNR BER slope still equals $P N_t$; if the slope is shallower than the rank bound predicts, the truncation in (12) is doing the diversity work rather than the physical channel.","tokens_in":24007,"feed_emoji":"📡","tokens_out":5762,"duration_ms":52499,"temperature":0.7,"pith_summary":"The paper sets out to give AFDM, a chirp-based waveform for high-mobility channels, a transmit-diversity capability by pairing cyclic delay diversity across several antennas with index modulation, which hides extra bits in which chirp subcarriers are switched on. It claims that if the normalized delays and Doppler shifts of all paths occupy disjoint slots in the discrete affine Fourier (DAF) domain, the two proposed schemes, CDD-AFDM-IM-I and CDD-AFDM-IM-II, achieve full diversity, and it states exactly when this happens as a counting inequality involving the number of subcarriers, paths, and antennas. The paper also derives closed-form bit-error-rate upper bounds whose slope in the high-SNR region predicts that diversity order, and it offers a low-complexity detector meant to approach that performance without exponential search. A sympathetic reader would care because a waveform that reliably separates delays and Dopplers with full antenna diversity is a candidate for next-generation links to fast-moving terminals.","feed_headline":"Chirp-waveform index modulation earns full transmit diversity","feed_subtitle":"The schemes pair cyclic delay diversity with index modulation on chirp subcarriers; a closed-form BER bound predicts the gain.","key_machinery":"The load-bearing object is the DAF-domain effective channel matrix $\\mathbf{H}_{\\epsilon,\\ell}$ of each antenna-path pair, whose non-zero entries sit on a band centered at $\\mathrm{loc}_{\\epsilon,\\ell}=-\\alpha_{\\epsilon,\\ell}+2N\\lambda_1(l_{\\epsilon,\\ell}+l_\\epsilon)$. Choosing the chirp parameter $\\lambda_1=(2\\alpha_{\\max}+1)/(2N)$ in the integer Doppler case, or $\\lambda_1=(2\\alpha_{\\max}+2k_\\alpha+1)/(2N)$ in the fractional case, and a cyclic delay interval $\\Delta_{\\min}=l_{\\max}+1$ between adjacent antennas pushes those bands apart, turning the non-overlap of the sets $A_{\\epsilon,\\ell}$ or $B_{\\epsilon,\\ell}$ into the full-diversity condition. Diversity itself is read from the rank of the difference matrix $\\Upsilon_{\\mathbf{x}_i}-\\Upsilon_{\\mathbf{x}_j}$: the minimum rank over distinct transmitted vectors is the diversity order, and the BER upper bound (31) is the pairwise-error-probability average that carries that rank to the error-rate curve. The double-layer message passing detector replaces exponential ML search by factor-graph messages that separate symbol likelihoods from activation-state constraints.","core_discovery":"The central claim is a set of parameter rules under which the proposed CDD-AFDM-IM schemes make every resolvable path in the effective channel matrix occupy its own non-zero band, so the matrix collecting all path responses has full column rank. For integer Doppler shifts the requirement is $(l_{\\max}+1)(2\\alpha_{\\max}+1)N_t \\le N$; for fractional Doppler shifts it becomes $(l_{\\max}+1)(2\\alpha_{\\max}+2k_\\alpha+1)N_t \\le N$, where $N$ is the number of chirp subcarriers, $l_{\\max}$ and $\\alpha_{\\max}$ are normalized delay and Doppler bounds, $N_t$ is the number of transmit antennas, and $k_\\alpha$ defines the assumed width of fractional-Doppler leakage. When this condition fails, the paper proves by a rank argument that the index bits still earn a diversity bonus: different activation patterns differ in at least two positions, so the diversity order of the index bits stays above that of ordinary modulated bits. The derived BER upper bound (31), built from the rank of the difference matrix, is what makes the diversity order visible before simulation.","pith_inferences":["Implicit in the analysis is that $k_\\alpha$, the assumed fractional-Doppler leakage width, governs a trade-off between diversity and spectral efficiency; making $k_\\alpha$ adapt to the actual Doppler spread could recover capacity when the channel has few fractional taps.","The same band-non-overlap logic could be re-applied to OTFS-style two-dimensional representations, where the equivalent condition would involve the product of delay and Doppler resolutions rather than a one-dimensional chirp index; if that analogy holds, the counting inequalities transfer to delay-Doppler grids.","Because the full-diversity condition is purely combinatorial in $N$, $l_{\\max}$, $\\alpha_{\\max}$, and $N_t$, it offers a testable design rule for other delay-Doppler waveforms that chirp a single time axis, including OCDM with an added cyclic delay.","The Jakes model used in simulations produces continuous Doppler values, so the reported fractional-Doppler gains depend on the unstated $k_\\alpha$ choice; a fairness benchmark should fix $k_\\alpha$ and report sensitivity."],"forward_implications":["When (21) or (22) holds, the CDD-AFDM-IM-I/II schemes reach the full diversity order $P N_t$, and the closed-form bound (31) predicts the high-SNR BER slope without Monte Carlo simulation.","When the path-antenna product exceeds $N$, the index bits retain a higher diversity order than the modulation bits, so the schemes degrade more gracefully than CDD-AFDM as antennas are added.","CDD-AFDM-IM-II, by forcing all groups to share one activation pattern, multiplies the number of non-zero entries in the difference vector by $g$ and therefore raises the ceiling on index-bit diversity.","The DLMP detector brings the complexity of large-$N$ systems down from exponential ML search to a polynomial iteration count, with only a few percent more floating-point operations than single-layer message passing but better BER.","In fractional Doppler channels, the same design works if the leakage is modelled as confined to $2k_\\alpha+1$ slots; increasing $k_\\alpha$ widens the guard bands and tightens condition (22)."],"supporting_citations":[{"why":"Supplies the DAFT input-output relation and the convention that fractional Doppler interference is confined to a band of width $2k_\\alpha+1$, used for the $\\lambda_1$ settings and condition (22).","marker":"[20]"},{"why":"Introduces AFDM as a full-diversity waveform for high mobility, the starting point this paper extends to multiple antennas and index modulation.","marker":"[18]"},{"why":"Companion conference paper on AFDM for next-generation networks, providing the DAF-domain representation and diversity background.","marker":"[19]"},{"why":"Earlier work on cyclic delay diversity with index modulation in OFDM, the technique this paper adapts to AFDM and linear time-varying channels.","marker":"[32]"},{"why":"The only prior cyclic delay-Doppler shift AFDM scheme, which lacked the parameter conditions, performance analysis, and detector that this paper adds.","marker":"[33]"},{"why":"Prior AFDM with index modulation and an MMSE detector, used as the baseline and the starting point for the IM design guidelines.","marker":"[30]"},{"why":"Block-wise index modulation and receiver design for OTFS, the source of the PEP/rank diversity derivation and the message-passing receiver concepts.","marker":"[11]"},{"why":"The single-layer message passing detector for OTFS whose complexity and BER the DLMP detector is compared against.","marker":"[39]"}],"fun_headline_variants":["Full diversity achieved for chirp index modulation","CDD-AFDM-IM: full diversity and a fast detector","New condition guarantees full diversity in AFDM-IM","Chirp-based index modulation earns full diversity","AFDM-IM: full diversity rules and low-complexity detection"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The analysis treats fractional Doppler interference as if it were confined to a band of width $2k_\\alpha+1$ around each path's nominal position, so the full-diversity conditions for fractional Doppler hold only if real leakage tails outside that band remain negligible, and the paper does not give the $k_\\alpha$ value used in its fractional-Doppler simulations.","fun_headline_variants_meta":{"raw":{"variants":["Full diversity achieved for chirp index modulation","CDD-AFDM-IM: full diversity and a fast detector","New condition guarantees full diversity in AFDM-IM","Chirp-based index modulation earns full diversity","AFDM-IM: full diversity rules and low-complexity detection"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000557,"raw_usage":{"total_tokens":2721,"prompt_tokens":1085,"completion_tokens":1636,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":701,"completion_tokens_details":{"reasoning_tokens":1555}},"tokens_in":701,"tokens_out":1636,"duration_ms":14120,"temperature":1.0,"reasoning_tokens":1555,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T20:09:42.358289+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a channel whose fractional Doppler shifts produce leakage outside the assumed $2k_\\alpha+1$ band (for example, two paths whose band centers satisfy (22) but whose true sinc-like tails overlap), simulate ML detection without truncating $\\mathbf{H}_{\\epsilon,\\ell}$, and check whether the high-SNR BER slope still equals $P N_t$; if the slope is shallower than the rank bound predicts, the truncation in (12) is doing the diversity work rather than the physical channel.","supporting_citations":[{"cited_title":"AFDM: A full di versity next generation waveform for high mobility communications ,","cited_arxiv_id":null,"evidence_quote":"Introduces AFDM as a full-diversity waveform for high mobility, the starting point this paper extends to multiple antennas and index modulation."},{"cited_title":"Affi ne frequency division multiplexing for next-generation wireless netwo rks,","cited_arxiv_id":null,"evidence_quote":"Companion conference paper on AFDM for next-generation networks, providing the DAF-domain representation and diversity background."},{"cited_title":"Cyclic delay diversi ty with index modulation for green internet of things,","cited_arxiv_id":null,"evidence_quote":"Earlier work on cyclic delay diversity with index modulation in OFDM, the technique this paper adapts to AFDM and linear time-varying channels."},{"cited_title":"Cyclic delay-doppler shift: A simple transmit diversity t echnique for delay-doppler waveforms in doubly selective channels,","cited_arxiv_id":null,"evidence_quote":"The only prior cyclic delay-Doppler shift AFDM scheme, which lacked the parameter conditions, performance analysis, and detector that this paper adds."},{"cited_title":"Affine Frequency Division Multiplexing With Index Modulation","cited_arxiv_id":"2310.05475","evidence_quote":"Prior AFDM with index modulation and an MMSE detector, used as the baseline and the starting point for the IM design guidelines."},{"cited_title":"Blo ck- wise index modulation and receiver design for high-mobilit y OTFS communications,","cited_arxiv_id":null,"evidence_quote":"Block-wise index modulation and receiver design for OTFS, the source of the PEP/rank diversity derivation and the message-passing receiver concepts."}],"review_version":1}