{"id":"63f59dd1-d87a-4f76-b562-21d98fed9b90","arxiv_id":"2502.08996","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Binary transmit masks with two-level periodic autocorrelation, such as m-sequences and Singer difference sets, make half-duplex ISAC range mainlobes constant and sidelobes near-ideal at duty cycles close to 50%.","lead":"Integrated sensing and communication systems that stop transmitting while they listen avoid self-interference but create range-blind gaps. This paper shows that carefully chosen on/off transmit patterns, built from classic cyclic difference-set sequences, keep those blind-zone effects small while supporting roughly 50% communication duty cycles.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Ideal-mask construction is internally sound; the load-bearing weakness is the transfer of the zero-mainlobe-fluctuation guarantee from the exact integer-delay, instant-switch model to physical half-duplex operation.","rationale":"The reader's weakest_assumption identifies the same weak point that I would stress: the ideal half-duplex model with exactly complementary masks, instantaneous switching, Nyquist pulses, and integer symbol-spaced delays. I checked the main algebraic claims and found no internal inconsistency in the ideal-mask construction: mainlobe ideality follows from the two-level ACF/CDS property, and the PESL-ideality proof rests on the integrality of the expected sidelobe entries plus the ceiling lower bound. The conditional verdict is therefore appropriate, and my stress test does not move it. The concrete test with fractional delay or receiver guard intervals would settle whether the practical viability claim survives when the model idealization is relaxed. If the degradation is negligible, the conditional can be upgraded; if it is large, the applicability claims need revision while the mathematical results stand.","tokens_in":22057,"tokens_out":20021,"duration_ms":192713,"concrete_test":"Use the N=63, rho=31/63 Singer PG(5,2) mask from Sec. VI and repeat the ARGI/PESL evaluation of Figs. 4 and 6 with a raised-cosine Nyquist pulse and target delays tau0=(k+epsilon)*Tc for epsilon in {0, 0.25, 0.5} and k in {1,...,N-1}, keeping all other settings identical. If the resulting mainlobe profile departs from the flat ideal-mask profile, or ARGI leaves zero for PSK symbols, the ideal-CDS guarantee does not transfer to non-sample-spaced targets. A companion run with a receiver guard interval of G in {0,1,2} symbols at mask transitions can separate the switching-transient cause.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Within the stated model, Proposition 7 and the supporting CDS arguments appear internally consistent: for Singer masks the mainlobe term [R]_{k,k} is constant across non-blind bins, and the PESL equals the ceiling of the AESL lower bound via the integrality argument in Corollary 1. The load-bearing step is therefore not the combinatorics but the model link in Sec. II-B: the derivation of z_k(n)=mr(n)mt(n-k)x_{n-k}, and hence every later ideal-mask result, requires exactly complementary instantaneous masks and an integer target delay tau0=k*Tc with Nyquist pulses. Outside this model the constant-mainlobe property is formally unprotected. The paper itself concedes in Sec. V that sample-level switching rates may be impractical, and the slow-time variant only moves the idealization to the slow-time mask. The headline claim that MASM is a viable high-throughput half-duplex ISAC waveform therefore rests on unquantified assumptions about switching transients/guard intervals and fractional-delay leakage; the simulations in Sec. VI do not exercise either. This is a conditional applicability gap, not an internal inconsistency.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes MASM, a half-duplex ISAC waveform scheme in which a periodic binary transmission mask selects symbol intervals for transmission and the complementary mask selects receive intervals. The central contributions are: (i) definitions of range glint intensity (RGI) and averaged range glint intensity (ARGI) as metrics for the delay-dependent mainlobe fluctuation of half-duplex sensing; (ii) closed-form expressions for the expected ARGI of conventional pulse radar, MASM with random masks, and MASM with data payload signals; (iii) a reduction of the mask-optimization problem to an ℓ4-norm minimization; (iv) a proof that masks derived from cyclic difference sets, in particular Singer CDSs, are simultaneously mainlobe-fluctuation-ideal and peak-expected-sidelobe-level (PESL) ideal in the fast-time model; and (v) an extension to slow-time coding with piecewise-constant masks. Numerical simulations verify the main analytical formulas.","tokens_in":22227,"tokens_out":24442,"duration_ms":226654,"significance":"If the results are correct, the paper offers a principled alternative to conventional pulse radar and PRF-staggering for half-duplex ISAC, achieving around 50% duty cycle while keeping the expected mainlobe level constant outside the k=0 blind bin for constant-modulus constellations. The fast-time ideal-mask construction is elegant: it connects a communication-centric ISAC waveform design problem to established cyclic difference set theory, and the derivations of Propositions 2, 3, 6, and 7 are carefully executed, with simulations reproducing the closed-form expressions. The paper is also honest in defining expectation-based metrics rather than claiming deterministic guarantees for random data payloads. However, the slow-time extension contains a load-bearing claim about PESL-ideality that is not supported and, as shown below, is false for a concrete parameter choice; additionally, the practical claims depend on an idealized instantaneous-switching, integer-delay model that is not quantified.","major_comments":[{"comment":"The sentence \"Singer CDSs are also ideal slow-time transmission masks in the sense detailed in Proposition 7\" is not supported by Corollary 2 and appears to be false. For a slow-time mask mt = fmt ⊗ 1_T with fmt a Singer mask of length L, equation (54) gives interior sidelobe levels equal to T times the corresponding slow-time levels, so for the PG(4,2) Singer mask (L=31, ρ=15/31) and T=16 we have N=LT=496 and PESL = T·q^{n-2} = 16·4 = 64. The universal lower bound of Corollary 1 for this (N,ρ) is ceil(60.05)=61, since AESL = 14,684,160/244,530 ≈ 60.05. Thus the slow-time mask does not achieve the PESL lower bound, contradicting the claimed PESL-ideality. The authors should either prove a slow-time-specific PESL bound or remove this claim.","section":"V-B, after Corollary 2"},{"comment":"The entire ideal-mask guarantee rests on the model z_k(n) = m_r(n) m_t(n-k) x_{n-k}, which requires exactly complementary instantaneous masks (m_r = 1 - m_t) and a target delay equal to an integer number of symbol intervals τ0 = kTc with Nyquist pulses. The paper itself concedes in Sec. V that sample-level switching may be impractical, and the slow-time variant still assumes no transition gaps and integer-delay echoes. No analysis or simulation is provided for fractional delays or switching transients, so the headline claim that MASM is a viable high-throughput half-duplex ISAC waveform is supported only inside this idealized model. The authors should either quantify the robustness of the constant-mainlobe and PESL properties under these non-idealities or explicitly restrict the applicability claims.","section":"II-B, equation for z_k(n)"},{"comment":"The ideal-mask guarantees for data payload signals are statements about the expected ARGI. For non-constant-modulus constellations (e.g., 16QAM/64QAM), the actual range glint is random, and the paper does not provide any concentration or worst-case bound; Fig. 9 shows only empirical averages over 2000 instances. Since high-throughput communication with QAM constellations is part of the advertised operating regime, the authors should either provide high-probability bounds on RGI for ideal masks or explicitly state that the zero-fluctuation property holds only for constant-modulus constellations and, for general constellations, only in expectation.","section":"III-C, Proposition 3 and Eq. (26)"}],"minor_comments":[{"comment":"The definition of the slow-time sidelobe level ea_{k,l} uses the summation limit N, but the slow-time mask has length L; the sum should run to L.","section":"Corollary 2"},{"comment":"The cyclic difference d_k is written as ea_{(k+1) mod N} - ea_k, but since ea is of length L, the modulus should be L, not N.","section":"Proposition 8"},{"comment":"The statement that Barker codes satisfy the \"two-level ACF\" condition is misleading: Barker codes have low aperiodic autocorrelation sidelobes, not a constant periodic autocorrelation sidelobe. The intended point about periodic two-level autocorrelation is correctly made by m-sequences and cyclic difference sets, so the Barker sentence should be revised or removed.","section":"III-C, paragraph on two-level ACF sequences"},{"comment":"In the proof of Proposition 8, the phrase \"for 1 < l < T\" should likely be \"for 1 ≤ l < T\" to cover the derived difference relation; the current wording omits l=1.","section":"Appendix III-A"},{"comment":"The caption contains a typo: \"Trapezoidal rm's suffer from severe mainlobe fluctuation\" should read \"Trapezoidal r_m suffers\" or \"Trapezoidal r_m vectors suffer\".","section":"Figure 3 caption"},{"comment":"The shift matrix eJ_k is defined for size 3N, but it is later used as an N×N shift matrix in Sec. II-C and elsewhere; the dimensions should be clarified to avoid confusion.","section":"Eq. (8)"}],"recommendation":"major_revision","confidential_remarks":"The fast-time results are strong and likely correct, but the slow-time PESL-ideality claim in Sec. V-B is contradicted by a direct calculation with the paper's own formulas (e.g., PG(4,2), T=16 gives PESL=64 versus lower bound 61). This is not a mere presentation issue; it changes the paper's contribution. The idealized switching/delay model is a separate, commonly accepted idealization, but given the paper's practical framing it should be addressed explicitly. I recommend major revision rather than rejection because the fast-time core is sound and the slow-time issue can be fixed by correcting the claim."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this is a genuinely useful analytic paper. The new metric (RGI/ARGI) is a clean way to talk about mainlobe fluctuation in half-duplex ISAC, and the observation that minimizing expected ARGI reduces to flattening the mask's power spectrum is a nice bridge to classical sequence design. The proof that Singer CDS masks are both mainlobe-fluctuation-ideal and PESL-ideal is the real contribution; it's grounded in classical difference-set theory, not circular, and the numerical results line up with the formulas. The authors deserve credit for the closed-form expected ARGI expressions (Propositions 2, 3, 6, 7) and for carrying the analysis through slow-time operation.\n\nThe soft spots are real but not fatal. The analysis is expectation-only: there are no concentration or worst-case bounds for the ARGI of a given random data payload, so a specific frame could show more mainlobe variation than the average suggests. The larger gap is the model link. All the ideal-mask results rely on exactly complementary transmit/receive masks with sample-level switching and on delays that land exactly on integer symbol intervals with Nyquist pulses. The paper itself acknowledges in Section V that sample-level switching may be impractical and shifts to slow-time masks, which only moves the idealization. The simulations don't exercise either of these issues, so the headline claim about ~50% duty cycle with mild mainlobe fluctuation is a conditional statement about an idealized system, not a demonstrated hardware-feasible one.\n\nThat said, the internal math is sound and the ideal construction is concrete and checkable. The paper would benefit from a section quantifying the effects of guard intervals or fractional delay, and from variance bounds, but the core contribution stands. I'd take it seriously: this deserves peer review rather than a desk rejection. It's most useful for people working on half-duplex ISAC waveform design or on the sensing side of 6G systems, and I'd probably cite it when I write about range blindness or mask design in ISAC.","headline":"Solid analytic paper on half-duplex ISAC mask design; the ideal-mask results are real, but practical claims outrun the idealized model.","tokens_in":22820,"tokens_out":2176,"would_cite":true,"duration_ms":20478,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["94A12","05B10","51E20"],"pacs":[],"model":"deepseek-v4-flash","headline":"A half-duplex ISAC waveform can carry data at roughly 50% duty cycle while keeping the expected echo mainlobe exactly flat across every non-blind range bin, if the transmission mask is built from a cyclic difference set.","keywords":["integrated sensing and communication","half-duplex waveform design","transmission mask","range glint intensity","mainlobe fluctuation","cyclic difference sets","finite projective geometry","peak expected sidelobe"],"falsifier":"Measure the expected mainlobe level over many constant-modulus data realizations for a finite-geometry ideal mask at, say, $\\rho\\approx 1/2$, while sweeping the target delay from $\\tau_0=kT_c$ to $\\tau_0=kT_c+\\epsilon$ for $0<\\epsilon<T_c$; if the variance of mainlobe levels across range bins grows noticeably with $\\epsilon$, the zero-ARGI claim is limited to the sample-spaced model. Equivalently, insert a one-sample guard interval between transmit and receive slots and check whether the expected ARGI remains zero; a nonzero value would show the exact complementarity assumption is essential.","tokens_in":21840,"feed_emoji":"📡","tokens_out":11942,"duration_ms":101733,"temperature":0.7,"pith_summary":"Integrated sensing and communication (ISAC) systems that transmit and receive simultaneously suffer self-interference, while classical half-duplex pulse radars avoid it only by using small duty cycles that limit communication throughput and create sizable blind ranges. This paper proposes MASked Modulation (MASM), a half-duplex scheme in which a periodic binary transmission mask marks which symbol slots transmit data, and the complementary mask marks which slots listen for echoes, so that a system can run near a 50% duty cycle without self-interference. The paper introduces a range-blindness metric called mainlobe fluctuation, defined through the variance of expected mainlobe levels across range bins, and shows that designing a good mask is equivalent to flattening the power spectrum of the mask. It proves by explicit construction that masks arising from cyclic difference sets built from finite projective geometries are ideal: zero averaged range glint for constant-modulus constellations and the minimum possible peak expected sidelobe. If the paper is right, half-duplex ISAC becomes a principled high-throughput waveform family with a clean tradeoff between communication throughput (growing with duty cycle) and sensing dynamic range (degrading with $1-\\rho$).","feed_headline":"Half-duplex ISAC keeps range flat at 50% duty cycle","feed_subtitle":"Masks built from cyclic difference sets give every non-blind range bin the same echo level while capping sidelobe peaks.","key_machinery":"The load-bearing object is the periodic binary transmission mask $\\mathbf{m}_t\\in\\{0,1\\}^N$ with duty cycle $\\rho=\\frac{1}{N}\\mathbf{1}^T\\mathbf{m}_t$, paired with the complementary reception mask $\\mathbf{m}_r=1-\\mathbf{m}_t$. All sensing metrics are functions of this mask: mainlobe levels are cross-correlations $a_k=(1-\\mathbf{m}_t)^T\\tilde{\\mathbf{J}}^k\\mathbf{m}_t$, the ARGI is their variance, and via the unitary DFT matrix $\\mathbf{F}$ it reduces to minimizing $\\|\\mathbf{F}\\mathbf{m}_t\\|_4^4$, i.e. flattening the mask's power spectrum. Ideal masks are exactly the binary sequences with two-level periodic autocorrelation, equivalently cyclic difference sets; the construction used for the strongest result is the cyclic incidence matrix of points and hyperplanes in a finite projective geometry, whose intersection structure fixes all mainlobes to $q^{n-1}$ and all peak sidelobes to $q^{n-2}$.","core_discovery":"The central claim is that range blindness in half-duplex ISAC can be removed, in expectation, by choosing transmission masks whose periodic autocorrelation has only two levels, and that the same masks can also achieve the ideal peak-sidelobe level. Concretely, for a mask $\\mathbf{m}_t\\in\\{0,1\\}^N$ with duty cycle $\\rho$, the expected mainlobe in range bin $k$ is $a_k=(1-\\mathbf{m}_t)^T\\tilde{\\mathbf{J}}^k\\mathbf{m}_t$; the averaged range glint intensity is the variance of these values over $k>0$, and after a discrete Fourier transform it reduces to a function of $\\|\\mathbf{F}\\mathbf{m}_t\\|_4^4$. A mask with a two-level autocorrelation makes all $a_k$ equal, giving exactly zero expected ARGI for constant-modulus data. The paper further shows that the lower bound on peak expected sidelobe is attained by the cyclic incidence matrices between points and hyperplanes in the finite projective space $\\mathrm{PG}(n,q)$, for which the duty cycle is $\\rho=(q^n-1)/(q^{n+1}-1)$, the common mainlobe is $q^{n-1}$, and the peak expected sidelobe is $q^{n-2}$; this family includes all $m$-sequences and some Barker codes. A slow-time extension using piecewise-constant masks keeps these ideal properties up to first order while supporting blind ranges comparable to the sub-pulse length.","pith_inferences":["If practical transceivers need guard intervals between transmit and receive slots, the exact complementarity $\\mathbf{m}_r=1-\\mathbf{m}_t$ is violated at boundaries; a natural extension would be to compute how ARGI and PESL degrade as a function of guard length and to design masks that remain near-ideal under the relaxed masks.","The equivalence between ideal masks and cyclic difference sets invites a broader search: other difference set families with density near $1/2$ could fill in $(N,\\rho)$ configurations the finite-geometry construction misses, and the PESL lower bound gives a quick screening test for candidates.","The sample-spaced Nyquist assumption means fractional target delays fall outside the proof; testing MASM with oversampled reception or with fractional-delay echoes would show whether the constant-mainlobe property survives inter-symbol interference across mask transitions.","Because the ideal masks are point-hyperplane incidence structures, they can be reinterpreted as hopping or scheduling patterns, which may transfer the same flat-mainlobe and low-peak-sidelobe guarantees to MIMO radar or spectral-spreading ISAC designs."],"forward_implications":["At duty cycle $\\rho=1/2$, the energy accumulation efficiency of a half-duplex ISAC system reaches its maximum of $1/4$, and ideal masks keep the expected mainlobe constant across all range bins $k>0$, so no range glint appears outside the single blind bin at $k=0$.","Because the average expected sidelobe level is independent of the mask, mask design can target the peak sidelobe; finite-geometry ideal masks attain the analytic PESL floor, so they leave no excess sidelobe spikes.","The asymptotic mainlobe-to-sidelobe ratio behaves as $(1-\\rho)N+\\mu_4+1$, which makes the throughput-dynamic-range tradeoff explicit: raising communication throughput by increasing $\\rho$ costs sensing dynamic range, and at a 50% duty cycle the cost is about 3 dB relative to a full-duplex scheme.","Slow-time piecewise-constant masks inherit the ideality of their fast-time counterparts, so the same cyclic difference set constructions apply at frame level, with blind range set by the sub-pulse length and throughput $\\rho LT$ symbols per PRI.","For generic constellations the expected ARGI is proportional to $\\mu_4-1$, so constant-modulus constellations are optimal for sensing; random masks have ARGI-to-mainlobe ratio $O(1/N)$, whereas conventional pulse masks have a constant ratio."],"supporting_citations":[{"why":"Supplies the finite-projective-geometry cyclic incidence construction that yields the simultaneously mainlobe-fluctuation-ideal and PESL-ideal masks.","marker":"[33]"},{"why":"Establishes that binary sequences with two-level autocorrelation are equivalent to cyclic difference sets, connecting ideal masks to the CDS formalism.","marker":"[29]"},{"why":"Supplies the OFDM lowest-ranging-sidelobe result and the full-duplex expected sidelobe level $N$ used as the benchmark in the comparison.","marker":"[11]"},{"why":"Defines conventional pulse-radar duty cycles and blind-range behavior that MASM is designed to improve upon.","marker":"[21]"},{"why":"Provides the classical two-level autocorrelation sequence framework used to identify mainlobe-fluctuation-ideal masks.","marker":"[28]"},{"why":"Documents the ripple and design problems of staggered-PRF radar that motivate the slow-time MASM comparison.","marker":"[25]"}],"fun_headline_variants":["Flat ISAC range at 50% duty from cyclic masks","Projective geometry flattens half-duplex ISAC range","Two-level autocorrelation kills range blindness in ISAC","Half-duplex ISAC: 50% duty, zero mainlobe ripple"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The analysis assumes that transmit and receive can switch perfectly between consecutive symbol slots (so the receive mask is the exact complement of the transmit mask), that the pulse shape does not leak into neighboring symbol slots, and that target echoes arrive exactly on symbol-boundary delays; if any of these fail in hardware, even an ideal mask may show mainlobe fluctuation.","fun_headline_variants_meta":{"raw":{"variants":["Flat ISAC range at 50% duty from cyclic masks","Projective geometry flattens half-duplex ISAC range","Two-level autocorrelation kills range blindness in ISAC","Half-duplex ISAC: 50% duty, zero mainlobe ripple"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000442,"raw_usage":{"total_tokens":2308,"prompt_tokens":1081,"completion_tokens":1227,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":697,"completion_tokens_details":{"reasoning_tokens":1154}},"tokens_in":697,"tokens_out":1227,"duration_ms":9953,"temperature":1.0,"reasoning_tokens":1154,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T23:00:17.597686+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the expected mainlobe level over many constant-modulus data realizations for a finite-geometry ideal mask at, say, $\\rho\\approx 1/2$, while sweeping the target delay from $\\tau_0=kT_c$ to $\\tau_0=kT_c+\\epsilon$ for $0<\\epsilon<T_c$; if the variance of mainlobe levels across range bins grows noticeably with $\\epsilon$, the zero-ARGI claim is limited to the sample-spaced model. Equivalently, insert a one-sample guard interval between transmit and receive slots and check whether the expected ARGI remains zero; a nonzero value would show the exact complementarity assumption is essential.","supporting_citations":[{"cited_title":"A theorem in finite projective geometry and some applications to number theory,","cited_arxiv_id":null,"evidence_quote":"Supplies the finite-projective-geometry cyclic incidence construction that yields the simultaneously mainlobe-fluctuation-ideal and PESL-ideal masks."},{"cited_title":"The merit factor of binary sequences related to difference sets,","cited_arxiv_id":null,"evidence_quote":"Establishes that binary sequences with two-level autocorrelation are equivalent to cyclic difference sets, connecting ideal masks to the CDS formalism."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Defines conventional pulse-radar duty cycles and blind-range behavior that MASM is designed to improve upon."},{"cited_title":"New nonbinary sequences with ideal two- level autocorrelation,","cited_arxiv_id":null,"evidence_quote":"Provides the classical two-level autocorrelation sequence framework used to identify mainlobe-fluctuation-ideal masks."},{"cited_title":"On the design of staggered moving target indicator filters,","cited_arxiv_id":null,"evidence_quote":"Documents the ripple and design problems of staggered-PRF radar that motivate the slow-time MASM comparison."}],"review_version":1}