{"id":"454a222f-1334-4534-95f9-8f34b165bedf","arxiv_id":"2411.08473","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"DA-FrFDM dynamically selects the fractional Fourier transform angle per symbol block to reduce PAPR while retaining simple one-tap equalization via a quadratic phase multiplication.","lead":"This paper proposes a multicarrier scheme that picks a per-block rotation angle in the fractional Fourier domain to lower peak-to-average power ratio, and adds a quadratic phase so the receiver can still equalize with one tap. The scheme reports PAPR gains over clipping, selective mapping, and partial transmit sequences for both QAM and Gaussian signals in simulations.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Algorithm 1's surrogate-based angle pruning is unproven; reported PAPR gains may not be attributable to the proposed search.","rationale":"The reader identified exactly the same load-bearing weakness: the surrogate I = ∫g(t)^4 is used to prune the angle search without any proof that its local minima correspond to true PAPR minima. I find this to be the single most consequential risk because it undermines the paper's distinctive mechanism—dynamic per-block angle selection via a low-complexity algorithm—rather than a peripheral detail. If the surrogate is unreliable, the reported PAPR reductions in Fig. 2 might still be real, but they would be attributable to the brute-force 128-evaluation budget rather than to the proposed search, invalidating the algorithmic contribution and the 'efficient' angle discovery claim. Conversely, if the surrogate works, the central PAPR claim is credible. The concrete test I propose directly measures whether Algorithm 1's candidate set contains near-optimal angles, and it is straightforward to run. I also note secondary issues (garbled proof of Theorem 4, possible typo in the oversampled phase term in Remark 2, and single-realization ICI plot), but these are repairable or auxiliary. The verdict remains CONDITIONAL because the surrogate concern is serious yet empirically checkable, not a proven fatal flaw.","tokens_in":12226,"tokens_out":24411,"duration_ms":214969,"concrete_test":"For N=64 with QAM and complex Gaussian symbols, generate at least 10^4 random blocks. For each block, run Algorithm 1 with a PAPR evaluation budget of 128 (as in the paper), and also find the globally best angle by evaluating true PAPR on a dense grid of at least 10^4 equally spaced angles in the Theorem 2 range. Compare the empirical CCDF of the PAPR achieved by Algorithm 1 against (a) the global optimum and (b) 128 randomly chosen angles. If the median gap to the global optimum exceeds 0.5 dB, or if Algorithm 1 is not significantly better than random 128-angle selection, then the surrogate assumption fails to support the claimed efficient discovery of a near-optimal angle.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The central claim that DA-FrFDM achieves significant PAPR reduction rests in large part on the practical viability of dynamically selecting the fractional angle per block. Algorithm 1 identifies candidate angles by detecting sign changes of the derivative of I = ∫g(t)^4 dt, relying on the implicit assumption that local minima of this L4 surrogate are located near angles that minimize the true max-based PAPR. The paper provides no proof, simulation, or ablation establishing this correspondence. If the surrogate's stationary points do not reliably contain near-optimal angles, the refinement in Algorithm 1 may systematically discard the angles that would actually minimize PAPR, leaving the candidate set no better than random or coarse-grid selection. The final step (evaluating true PAPR over Ω) only selects among the surrogate-filtered candidates; it cannot recover angles that the surrogate has already discarded. Since the claimed large PAPR reduction (Fig. 2) is generated using this algorithm, the load-bearing assumption is that the surrogate-guided pruning preserves the low-PAPR angles. That assumption is unverified, and the paper contains no control experiment comparing Algorithm 1 with an exhaustive search or with random selection under the same 128-evaluation budget. This is a correctness risk in the core mechanism, not merely a complexity concern.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes dynamic-angle fractional Fourier division multiplexing (DA-FrFDM), in which each data block is transformed by an inverse discrete fractional Fourier transform (IDFrFT) whose angle α is chosen per block to reduce PAPR. A quadratic phase term is applied in the time domain to enable one-tap equalization with an angle-independent channel response. The paper derives a closed-form envelope-power expression (Sec. III-A), proves a periodicity property of PAPR in the transformed angle variable (Theorem 2), and proposes a derivative-based surrogate search (Algorithm 1) to identify candidate angles. Simulations in Sec. V report PAPR at CCDF 10^-3 roughly half the OFDM baseline and better than clipping, SLM, and PTS for both complex Gaussian and QAM signals, with comparable BER/MSE in Rayleigh fading and a favorable PAPR-ICI trade-off in a doubly dispersive channel.","tokens_in":12430,"tokens_out":6703,"duration_ms":68323,"significance":"If the claims hold, DA-FrFDM provides a PAPR-reduction method that avoids reserved tones and phase-sequence side information, applies to continuous-amplitude Gaussian signals, and retains simple equalization and ICI mitigation. The analytical pieces are genuinely useful: Eq. (5) gives a compact expression of envelope power as a function of α through Aα; Theorem 2 restricts the search interval; Theorem 3 proposes a structured computation of the derivative of a surrogate cost. The simulation protocol of fixing a PAPR-evaluation budget of 128 for all compared methods is a fair and commendable feature. However, the two main advertised contributions—efficient discovery of the optimal angle and α-independent one-tap equalization—rest on arguments that are currently incomplete, so the paper needs substantial revision before the claims can be accepted.","major_comments":[{"comment":"The surrogate-based search is not validated. Algorithm 1 prunes the search space by detecting sign changes of the derivative of I = ∫₀ᵀ g(t)⁴ dt, then evaluates the true PAPR only over the surviving set Ω. The paper neither proves nor experimentally tests that local minima of I are reliable proxies for angles that minimize the true max-based PAPR η(Aα). Since |x(t)|² = (1/N)Σ|s[k]|² + (2/N)g(t), the peak of |x(t)|² is determined by max g(t), whereas ∫g⁴ is an L4 norm over the whole interval; minimizing it can reduce energy away from the peak without reducing the maximum. Moreover, the coarse initial grid and the local sign-change condition can miss local minima of I itself, and the final argmin over Ω cannot recover angles already discarded. A control experiment comparing Algorithm 1 with an exhaustive fine search or with random candidate selection under the same 128-evaluation budget is needed to establish that the surrogate pruning preserves low-PAPR angles. Without such evidence, the claimed efficient discovery of α* is unsupported, even if the final PAPR reduction itself may be real.","section":"Section III-B, Eq. (7)-(9), Algorithm 1"},{"comment":"The proof of Theorem 4 is garbled and does not establish the claimed α-independent one-tap equalization. The symbol ex[n] appears without definition, the phase term e^{j/2 n² cot α (T_s L)²} is introduced without justification, and the change of variables that combines the two circular-convolution sums does not correctly track the quadratic phase factors, especially for the term with n − m + N. The notation hf = F1 h is not defined (a DFT is presumably intended, but this must be stated). In addition, the phase function θ(n) = 1/2 n² cot α T_s² is dimensionally inconsistent as written, since T_s² has units of time squared. Because one-tap equalization is one of the three advertised advantages, this theorem and proof must be rewritten carefully and verified before the equalization claim can be accepted.","section":"Section IV, Theorem 4 and its proof"},{"comment":"The signaling overhead of the dynamic angle is not quantified. DA-FrFDM selects a different α* for each block and must convey it to the receiver; the paper states that overhead is minimal but does not specify the number of bits per block, the effect of angle quantization on PAPR, or the resulting loss of spectral efficiency. The comparison in Fig. 2 fixes the PAPR-evaluation budget for all methods, which is good, but SLM and PTS also require side information, and a fair comparison should state the total overhead (in bits per block) for each scheme. This is load-bearing for the claim that DA-FrFDM achieves PAPR reduction with minimal signaling overhead.","section":"Sections II and V-A"}],"minor_comments":[{"comment":"The step sizes are written as '∆α1' and '∆α1/39'; these should read '∆α' and '∆α'/39' for consistency with Algorithm 1, and the ratio ∆α/∆α' should be stated to be an integer.","section":"Table I"},{"comment":"The continuous-time signal in Eq. (4) uses t², while the discrete version in Eq. (1) uses n²T_s²; the normalization of time units should be clarified so that the phase exponents are dimensionless.","section":"Equations (1) and (4)"},{"comment":"The set builder {α_i + j∆α′ : j = 0, 1, 2, ..., ∆α/∆α′} should explicitly restrict j so that the candidate values lie within the search interval from Theorem 2.","section":"Algorithm 1, line 10"},{"comment":"The CCDF subcaptions contain garbled symbols (e.g., 'Γୢ୆' and 'Prηୢ୆ > Γୢ୆'); these should be replaced with standard notation such as Pr(η > Γ).","section":"Figure 2 captions"},{"comment":"The system is referred to as 'DA-DFrFT' instead of 'DA-FrFDM'; the terminology should be consistent throughout.","section":"Remark 1"},{"comment":"The range statement assumes T²/π ≤ 1 for sin⁻¹ to be defined; the paper should either state this condition or handle the case T²/π > 1 explicitly, since the simulations use a specific value of T.","section":"Theorem 2"}],"recommendation":"major_revision","confidential_remarks":"The central idea is promising and the empirical PAPR reduction is worth pursuing, but the surrogate-search validation and the equalization proof must be fixed. If the authors can provide a control experiment for Algorithm 1 and a clean proof of Theorem 4, the paper would be suitable for publication. Without those, the two headlined advantages ('efficient optimal-angle discovery' and 'one-tap equalization') are not established."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The new thing here is real: per-block dynamic selection of the fractional Fourier angle for PAPR reduction, paired with a quadratic phase term that keeps one-tap equalization in the fractional domain. That combination is not in the cited prior art, and the equalization trick is a legitimate contribution. The PAPR reduction itself is partly by construction—the algorithm evaluates actual PAPR and picks the best angle, so the gains in Fig. 2 are not a fitted prediction; they are the output of a search. That is the same logic as SLM/PTS and is fine.\n\nThe soft spot is Algorithm 1. The paper prunes the search space using local minima of I = ∫g(t)^4 dt, a surrogate for max g(t), but never proves or demonstrates that the surrogate's minima contain the true PAPR-minimizing angles. The final step can only select among the surrogate-filtered candidates. No control experiment compares Algorithm 1 with exhaustive search or random selection under the 128-evaluation budget. If the surrogate systematically discards good angles, the reported gains are not attributable to the proposed search. This is the main correctness risk, and it is about the efficiency claim rather than the feasibility of dynamic-angle PAPR reduction. Even a dense grid over alpha would likely show gains, so the concept survives; the algorithm as presented does not.\n\nOther issues are proportional: Theorem 4's proof is garbled (the exponent variables don't cancel cleanly as written), the doubly dispersive ICI section rests on a single illustrative curve, and there is no code/data or error bars. These are repairable. The citations look honest; the FrFT-OFDM prior art is there.\n\nVerdict: worth a serious referee. I would send it out, but with a clear request to fix Theorem 4, justify the surrogate (formally or with an ablation), and add a control experiment for Algorithm 1. I wouldn't cite it yet in my own work, but I'd bring it to the group to see if others think the dynamic-angle idea generalizes.","headline":"Genuinely new dynamic-angle FrFT-PAPR scheme with a clever equalization trick, but the angle-search algorithm rests on an unproven surrogate and the ICI evidence is thin.","tokens_in":12988,"tokens_out":3499,"would_cite":false,"duration_ms":32980,"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":"This paper claims that per-block selection of a fractional Fourier angle can cut the peak-to-average power ratio of multi-carrier signals to nearly half the OFDM baseline, while preserving one-tap equalization and ICI mitigation.","keywords":["fractional Fourier transform","PAPR reduction","OFDM","DFrFT","one-tap equalization","doubly dispersive channels","QAM","Gaussian signals"],"falsifier":"Run Algorithm 1 on a large set of random QAM and Gaussian blocks with the same parameters as reported in the paper, then exhaustively evaluate true PAPR over the full angle range with a very fine grid. If any block's true PAPR-minimizing angle lies outside Algorithm 1's candidate set, or if exhaustive search yields a PAPR at least 0.5 dB lower than Algorithm 1's chosen angle, then the surrogate-guided search is not reliable as stated.","tokens_in":11972,"feed_emoji":"📡","tokens_out":5414,"duration_ms":50781,"temperature":0.7,"pith_summary":"This paper introduces DA-FrFDM, a multi-carrier modulation scheme that reduces peak-to-average power ratio (PAPR) by transmitting each data block in a fractional Fourier domain whose angle is dynamically chosen per block. The paper argues that PAPR behaves in a dual manner between the time and frequency domains, and an intermediate fractional angle can balance the two, cutting the PAPR at a CCDF of $10^{-3}$ to roughly half of the OFDM baseline for both QAM and Gaussian symbols. It further shows that a simple quadratic phase sequence makes the received signal equalizable with one tap, so the PAPR gain does not come at the cost of receiver complexity. This matters because high PAPR forces inefficient power-amplifier back-off, and the scheme also suits the Gaussian-like signals found in newer communication paradigms such as joint source-channel coding.","feed_headline":"Dynamic fractional angle cuts OFDM peak power by about half","feed_subtitle":"Choosing a fractional Fourier angle per block trims peaks and keeps one-tap equalization.","key_machinery":"The central object is the discrete fractional Fourier transform (DFrFT) with angle $\\alpha\\in(0,\\pi)$, whose inverse maps data symbols placed in the fractional domain to time-domain samples. PAPR is expressed as a periodic function of the combined parameter $A_\\alpha$, and the derivative of the surrogate $I=\\int_0^T g(t)^4\\,dt$ is evaluated by a trigonometric product-integral identity (Theorem 3) that reduces the computation to checking which frequency terms coincide. Algorithm 1 uses sign changes of $I'(\\alpha)$ to build a short candidate set, then evaluates true PAPR only over that set. The last piece is the quadratic phase sequence $\\theta(n)=\\frac{1}{2}n^2\\cot\\alpha\\,T_s^2$, which makes a circular convolution theorem hold in the ordinary frequency domain, so the channel response does not need to be recomputed when $\\alpha$ changes.","core_discovery":"The central claim is that dynamic per-block selection of the fractional Fourier transform angle can serve as a standalone PAPR reduction mechanism that outperforms clipping, selective mapping, and partial transmit sequence, without requiring extra spectrum or heavy side information. The argument rests on three results: PAPR depends on the angle only through the parameter $A_\\alpha = -\\frac{\\pi^2\\sin(2\\alpha)}{T^2}$; PAPR is $\\pi$-periodic in $A_\\alpha$, so the whole search collapses into a narrow angle interval; and the derivative of a surrogate integral $I = \\int_0^T g(t)^4\\,dt$ can be computed in closed form, letting Algorithm 1 isolate promising angles cheaply. Theorem 4 then proves that multiplying time samples by a quadratic phase sequence $e^{j\\theta(n)}$ with $\\theta(n)=\\frac{1}{2}n^2\\cot\\alpha\\,T_s^2$ diagonalizes the channel in the fixed frequency domain, so equalization stays a single tap regardless of $\\alpha$. The paper demonstrates with simulations that the PAPR gain persists for both QAM and complex Gaussian inputs, and that the ICI penalty of the angle search is only about 0.0004 dB in the doubly dispersive channel example.","pith_inferences":["If the surrogate derivative reliably locates the true optimum, the same bracket-and-refine search could be adapted to other peak metrics, such as the cubic metric, which matters for some amplifier models.","The per-block angle is side information the receiver must know; a natural testable extension is to estimate $\\alpha$ at the receiver from a small pilot block, or to restrict the angle set to a codebook to avoid explicit signaling.","In MIMO or OFDMA downlinks, assigning different angles to different users would break subcarrier orthogonality; an extension is to use one common angle per resource block and study whether PAPR gains survive scheduling constraints.","The nearly negligible ICI cost suggests a nearly orthogonal trade-off dimension, so a joint PAPR-and-ICI optimization over $\\alpha$ could yield a two-dimensional operating curve for doubly dispersive channels."],"forward_implications":["DA-FrFDM can be dropped into existing OFDM-style transceivers by replacing the FFT with an inverse DFrFT and adding a per-block angle search, keeping the same subcarrier structure.","Systems using continuous-amplitude symbols, such as joint source-channel coding or analog transmission, get PAPR relief without switching to a different waveform.","Because equalization remains one tap per subcarrier, receivers keep the low complexity of OFDM while gaining a lower peak factor.","The angle $\\alpha$ becomes a controllable knob for trading PAPR against inter-carrier interference in fast-fading channels, with the example showing a PAPR cut at a cost of only 0.0004 dB of interference power.","Lower PAPR can translate directly into higher power-amplifier efficiency or relaxed linearity requirements, which is economically relevant in base stations and user devices."],"supporting_citations":[{"why":"Provides the closed-form discrete fractional Fourier transform and affine transform expressions used to write the DA-FrFDM signal model in Equation (1).","marker":"[18]"},{"why":"Defines the discrete fractional Fourier transform (DFrFT) that underlies the whole scheme and its properties.","marker":"[17]"},{"why":"Surveys PAPR reduction techniques and sets the baseline problem statement that motivates the new method.","marker":"[2]"},{"why":"Justifies approximating the max operation by an $n$-th root integral, which is the basis for the surrogate function $I=\\int g(t)^4\\,dt$ used in Algorithm 1.","marker":"[20]"},{"why":"Selective mapping (SLM) is one of the three baseline PAPR reduction schemes that DA-FrFDM is compared against in Fig. 2.","marker":"[12]"},{"why":"Partial transmit sequence (PTS) is another baseline scheme compared in Fig. 2 under an equal PAPR evaluation budget.","marker":"[13]"},{"why":"Clipping and filtering is the third baseline scheme, shown in Fig. 2 and Fig. 3 for PAPR and decoding comparison.","marker":"[11]"},{"why":"Establishes the connection between the fractional Fourier transform angle and inter-carrier interference in doubly dispersive channels, which supports the ICI discussion in Section V-C.","marker":"[22]"},{"why":"Illustrates that existing FrFT-based systems require complex channel manipulation in the fractional domain, which Theorem 4 of this paper avoids via a fixed frequency-domain channel response.","marker":"[19]"}],"fun_headline_variants":["Per-block Fourier angle trims OFDM peaks, no side info needed","Angle search beats clipping and SLM for peak power reduction","Fractional Fourier trick halves PAPR, keeps one-tap equalization","Dynamic angle cuts PAPR, outperforms standard OFDM methods","Quadratic phase sequence slashes peak power, preserves ICI mitigation"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The entire efficient search relies on the belief that the local minima of the auxiliary quantity $I=\\int_0^T g(t)^4\\,dt$ are good stand-ins for the angles that actually minimize true peak power; if that link breaks for some data blocks, Algorithm 1 can miss the true best angle even though an exhaustive search would still find it.","fun_headline_variants_meta":{"raw":{"variants":["Per-block Fourier angle trims OFDM peaks, no side info needed","Angle search beats clipping and SLM for peak power reduction","Fractional Fourier trick halves PAPR, keeps one-tap equalization","Dynamic angle cuts PAPR, outperforms standard OFDM methods","Quadratic phase sequence slashes peak power, preserves ICI mitigation"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000199,"raw_usage":{"total_tokens":1369,"prompt_tokens":938,"completion_tokens":431,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":554,"completion_tokens_details":{"reasoning_tokens":341}},"tokens_in":554,"tokens_out":431,"duration_ms":4968,"temperature":1.0,"reasoning_tokens":341,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T21:32:42.466943+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run Algorithm 1 on a large set of random QAM and Gaussian blocks with the same parameters as reported in the paper, then exhaustively evaluate true PAPR over the full angle range with a very fine grid. If any block's true PAPR-minimizing angle lies outside Algorithm 1's candidate set, or if exhaustive search yields a PAPR at least 0.5 dB lower than Algorithm 1's chosen angle, then the surrogate-guided search is not reliable as stated.","supporting_citations":[{"cited_title":"Closed-form discrete fractional and affine Fourier transforms,","cited_arxiv_id":null,"evidence_quote":"Provides the closed-form discrete fractional Fourier transform and affine transform expressions used to write the DA-FrFDM signal model in Equation (1)."},{"cited_title":"Peak-to-average power ratio reduction in OFDM systems: A survey and taxonomy,","cited_arxiv_id":null,"evidence_quote":"Surveys PAPR reduction techniques and sets the baseline problem statement that motivates the new method."},{"cited_title":"A closed form selected mapping algorithm for PAPR reduction in OFDM multicarrier transmission,","cited_arxiv_id":null,"evidence_quote":"Justifies approximating the max operation by an $n$-th root integral, which is the basis for the surrogate function $I=\\int g(t)^4\\,dt$ used in Algorithm 1."},{"cited_title":"A comparison of peak power reduction schemes for OFDM,","cited_arxiv_id":null,"evidence_quote":"Selective mapping (SLM) is one of the three baseline PAPR reduction schemes that DA-FrFDM is compared against in Fig. 2."},{"cited_title":"Peak-to-average power ratio reduction of an OFDM signal using partial transmit sequences,","cited_arxiv_id":null,"evidence_quote":"Partial transmit sequence (PTS) is another baseline scheme compared in Fig. 2 under an equal PAPR evaluation budget."},{"cited_title":"Effects of clipping and filtering on the performance of OFDM,","cited_arxiv_id":null,"evidence_quote":"Clipping and filtering is the third baseline scheme, shown in Fig. 2 and Fig. 3 for PAPR and decoding comparison."},{"cited_title":"Near-optimal angle of transform in FRFT-OFDM systems based on ICI analysis,","cited_arxiv_id":null,"evidence_quote":"Establishes the connection between the fractional Fourier transform angle and inter-carrier interference in doubly dispersive channels, which supports the ICI discussion in Section V-C."},{"cited_title":"A new affine fractional Fourier transform with circular convolution property,","cited_arxiv_id":null,"evidence_quote":"Illustrates that existing FrFT-based systems require complex channel manipulation in the fractional domain, which Theorem 4 of this paper avoids via a fixed frequency-domain channel response."}],"review_version":1}