{"id":"27fff1db-57b2-4851-9c7a-571dd2319afa","arxiv_id":"2508.06540","paper_version":1,"verdict":"UNVERDICTED","confidence":"LOW","novelty_score":6.0,"correctness_risk":"unknown","formal_verification":"none","parameter_count":0,"one_line_summary":"Two AMP-based algorithms jointly detect active devices and estimate channels in OFDM-based grant-free wideband access, with state-evolution analysis and improved convergence.","lead":"This paper proposes two new signal-processing algorithms, AMP-A-EC and AMP-A-AC, that detect which devices are active and estimate their channels in wideband OFDM grant-free access systems. The methods aim to make such detection more accurate and faster than existing approaches, especially when the number of active devices is large relative to the pilot length.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Abstract alone cannot support the performance claim; the load-bearing unverified link is whether AMP-A-EC/AC actually converge to the MAP/MMSE solutions under the stated wideband model.","rationale":"I read the abstract in good faith and identified the strongest claim: the two AMP algorithms approximately solve MAP activity detection and MMSE channel estimation, with one analyzed by state evolution and the other offering lower complexity, and numerical results showing superior performance. The reader's weakest assumption was that the factor graph exactly captures channel and activity statistics. I agree that this is a necessary condition, but it is not the most load-bearing one: even if the factor graph is exact, the AMP updates need to converge to the posterior quantities of interest, and the state evolution needs to predict that convergence. The abstract itself flags convergence difficulty in the short-pilot regime but only says the algorithms alleviate it. Without the derivation and numerical evidence, the central performance claim is unverified. This is consistent with the reader's UNVERDICTED verdict, so I do not change the verdict. However, I would frame the concern more precisely as an unverified algorithmic convergence claim rather than primarily a model misspecification concern, hence 'partial' agreement with the reader's weakest assumption. The proposed test—reproducing state evolution and comparing against Monte Carlo in the short-pilot regime—would directly settle whether the central claim holds in the regime the authors target.","tokens_in":809,"tokens_out":1913,"duration_ms":23181,"concrete_test":"Obtain the full manuscript and independently reproduce the state evolution recursions for AMP-A-EC (the section analyzing error probability and MSE), then compare their predicted values against Monte Carlo simulations of the actual AMP-A-EC and AMP-A-AC algorithms on OFDM channels with the same hyperparameters. Focus on the regime where the pilot length is smaller than or comparable to the number of active devices. If the empirical MSE or activity-detection error probability deviates systematically from the state evolution prediction beyond Monte Carlo error, the claim that the algorithms approximately solve the MAP/MMSE problems is not supported in that regime.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is that AMP-A-EC and AMP-A-AC approximately solve the MAP activity detection and MMSE channel estimation problems, with state evolution analysis for AMP-A-EC and lower dominant-term complexity for AMP-A-AC, and that both outperform existing methods. For this claim to hold, two conditions must be true: (1) the factor graph exactly captures the true statistics of time-domain channels and device activities, and (2) the AMP updates converge to the relevant posterior quantities in the finite-dimensional OFDM regime, with state evolution accurately predicting their error probability and MSE. The abstract explicitly concedes that AMP has an inherent convergence problem when the pilot length is smaller than or comparable to the number of active devices, and claims the proposed algorithms only 'alleviate' it, not eliminate it. Because the full derivation, state evolution equations, and numerical comparisons are not available in this abstract-only submission, the performance and 'preferable regions' claims cannot be checked. This is not an internal inconsistency, but a missing-evidence concern: if the state evolution does not track actual algorithm behavior in the short-pilot regime, the claimed superiority over existing methods may not hold. The concern is load-bearing because the entire contribution rests on the algorithms approximately solving the stated MAP/MMSE problems, not merely on the model being exact.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper addresses joint device activity detection and channel estimation for OFDM-based grant-free access in wideband systems under frequency-selective fading. It proposes an exact time-domain signal model, formulates a MAP-based activity detection problem and two MMSE-based channel estimation problems, constructs a factor graph capturing the true statistics, and derives two AMP-based algorithms, AMP-A-EC and AMP-A-AC. The abstract claims that AMP-A-EC is analyzed via state evolution, that AMP-A-AC has lower dominant-term computational complexity, and that numerical results show superior performance with distinct preferable regions for the two algorithms.","tokens_in":1071,"tokens_out":2887,"duration_ms":35432,"significance":"If the claimed results are fully substantiated, the work would be practically relevant for massive grant-free access in wideband OFDM systems, especially the exact time-domain model and the new MAP/MMSE formulations. The paper does not appear to rely on fitted parameters; the analytical claims rest on standard model assumptions. However, the available material is only the abstract, so the significance cannot be confirmed without examining the full derivations, state-evolution equations, and numerical evidence.","major_comments":[{"comment":"The central claim that AMP-A-EC and AMP-A-AC approximately solve the stated MAP/MMSE problems is unsupported in the provided text. No derivations, update equations, or convergence conditions are shown. The authors should provide the factor-graph construction, the AMP update equations, and a precise statement of the conditions under which these algorithms converge to the MAP/MMSE solutions.","section":"Abstract"},{"comment":"The claimed state-evolution analysis for AMP-A-EC is asserted but not presented. The authors should give the state-evolution recursion, state its assumptions (e.g., large-system limit, pilot statistics), and demonstrate that its predictions match finite-dimensional simulations for the OFDM wideband model.","section":"Abstract"},{"comment":"The complexity claim that AMP-A-AC has 'lower computational complexity (in dominant term)' lacks a baseline and a derivation. The authors should specify the reference algorithm, provide per-iteration complexity expressions, and state how the dominant term is obtained.","section":"Abstract"},{"comment":"The numerical results are summarized only as 'superior performance and respective preferable regions.' The authors should report the simulation setup, baseline methods, performance metrics (e.g., activity detection error probability, channel estimation MSE), and define the parameter regions in which each algorithm is preferred.","section":"Abstract"}],"minor_comments":[{"comment":"The acronyms EC and AC in the algorithm names AMP-A-EC and AMP-A-AC are not defined.","section":"Abstract"},{"comment":"The phrase 'preferable regions' is vague; specifying whether these are regions in the pilot-length/active-device-number plane or some other parameter space would improve clarity.","section":"Abstract"},{"comment":"The term 'massive grant-free access' could be made more precise by giving example numbers for the number of devices, the number of active devices, and the pilot length.","section":"Abstract"}],"recommendation":"major_revision","confidential_remarks":"The review was conducted on the abstract alone; the full text is necessary to verify the algorithmic derivations, state-evolution analysis, complexity comparison, and numerical claims. Given the load-bearing nature of these missing pieces, I recommend major revision or, alternatively, that the editor obtain the full manuscript before a final decision. No citation-pattern or novelty concerns are visible from the abstract."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: this abstract describes a real, if incremental, piece of work in the AMP-for-grant-free-access line. The authors claim a new exact time-domain signal model, first-time formulations for the MAP activity detection problem and one MMSE channel estimation problem, a new factor graph, and two AMP algorithms with state evolution (for one) and complexity analysis (for the other). Those are concrete, checkable contributions. If the derivations and simulations hold, the paper would be useful to people working on wideband massive IoT, so the subfield-level significance is fair. Credit where due: the abstract is honest that the algorithms only 'alleviate' AMP's convergence problem in the short-pilot regime, not solve it, and the claimed preferable regions temper any universal superiority claim. That is good scientific framing.\n\nThe soft spots are all about what we cannot see. We have no equations, no state evolution details, no numerical results, and no reference list. The reader's low-confidence unverdict is exactly right; this is not an indictment of the paper, just an acknowledgment that the central performance claims cannot be checked from the abstract. The stress-test note worries that the factor graph must exactly capture the true statistics for the MAP/MMSE objectives to be well specified. That premise is standard in this literature, and the authors explicitly derive from an 'exact' time-domain model, so the model specification is less of a concern than the missing convergence evidence. The bigger real risk is the usual AMP gap: state evolution predicts large-system behavior, but the paper claims improved practical performance in the short-pilot regime where finite-dimensional effects matter. That is exactly where a referee should push for simulation corroboration.\n\nBottom line: this looks like a serious technical paper that deserves a rigorous peer review. It does not deserve a desk reject, and it does not deserve a pass based on an abstract. I would not cite it until I have read the full text, but I would bring it to a reading group to see whether the algorithms hold up.\n\nRecommendation: send to peer review with the expectation that referees check the derivations and, most importantly, whether the simulation results actually track the state evolution predictions in the short-pilot regime.","headline":"A plausible AMP-based contribution to OFDM grant-free access, but the abstract alone leaves all load-bearing claims unverifiable; worth a full-text referee.","tokens_in":1524,"tokens_out":1274,"would_cite":false,"duration_ms":17924,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["94A12","94A13"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper claims that two approximate message passing algorithms, AMP-A-EC and AMP-A-AC, jointly approximate the optimal activity detection and channel estimation for OFDM-based grant-free wideband access under frequency-selective fading.","keywords":["approximate message passing","device activity detection","channel estimation","grant-free access","OFDM","wideband systems","state evolution","factor graph"],"falsifier":"Simulate grant-free access with channel taps drawn from a distribution that differs from the prior assumed in the factor graph (for example, correlated taps or non-Gaussian fading), then compare AMP-A-EC's empirical detection error probability and MSE with its state-evolution prediction. A systematic gap that grows with the mismatch would show the state-evolution analysis and the algorithm's optimality claim depend on the exact prior.","tokens_in":662,"feed_emoji":"📡","tokens_out":4298,"duration_ms":46590,"temperature":0.7,"pith_summary":"This paper tackles the task of identifying which devices are active and estimating their channels in OFDM-based grant-free access, a setting for wideband systems where many devices can transmit without prior scheduling. The authors build an exact time-domain signal model under frequency-selective fading and derive the maximum a posteriori activity detection problem and two minimum mean square error channel estimation problems, with the MAP detection and one MMSE problem formulated for the first time. They then construct a factor graph that captures the exact statistics of time-domain channels and device activities, and propose two approximate message passing algorithms, AMP-A-EC and AMP-A-AC, that approximate the solutions to these problems. The paper shows that both algorithms avoid the convergence failure that plagues standard AMP when the pilot length is comparable to or smaller than the number of active devices, and that one of them has provably lower dominant-term complexity. Numerical results indicate each algorithm is preferable in a different operating region, which matters for making grant-free wideband access practical.","feed_headline":"Two AMP algorithms lift grant-free OFDM accuracy or speed","feed_subtitle":"They jointly find active devices and estimate channels in wideband systems, and each wins in a different regime.","key_machinery":"The central object is a factor graph that exactly captures the statistics of the time-domain channels and device activities. This factor graph encodes the joint posterior distribution used in the MAP activity detection and MMSE channel estimation objectives. The approximate message passing algorithms iterate on this graph to approximate the desired estimates; the state evolution analysis of AMP-A-EC is the analytic tool that predicts its detection error probability and channel estimation MSE.","core_discovery":"The paper's claim is that two AMP-based algorithms, AMP-A-EC and AMP-A-AC, approximately solve the MAP-based device activity detection problem and two MMSE-based channel estimation problems for OFDM-based grant-free wideband access under frequency-selective fading. AMP-A-EC is analyzed through state evolution, yielding predicted error probability and mean square error, while AMP-A-AC is shown to have lower computational complexity in the dominant term. Both algorithms fix AMP's known convergence problem when pilot length is small or comparable to the number of active devices. The paper further claims numerical evidence that both algorithms outperform existing methods in accuracy or computation time, with different preferable regions for each.","pith_inferences":["A natural extension the authors do not explore is to adapt these algorithms to time-varying channels or to multi-antenna base stations; the factor-graph formulation would need an added dimension for antennas or Doppler, but the same message-passing structure could carry over.","The state-evolution result suggests a practical test: one could use AMP-A-EC's predicted error probability to compute the smallest pilot length that meets a target detection reliability, a quantity not explicitly derived in the paper.","Because both algorithms are derived from a factor graph that assumes a particular prior, their performance under real measured channel statistics is an empirical question the paper does not settle; applying them to field data would be a direct test of the model-mismatch limit."],"forward_implications":["If the claims hold, OFDM-based grant-free wideband systems can jointly detect active devices and estimate frequency-selective channels without first scheduling users, using algorithms that are both accurate and fast enough for short pilot lengths.","The state-evolution analysis for AMP-A-EC gives a closed-form prediction of detection error probability and channel estimation MSE, so system designers can tune pilot length and thresholds analytically rather than by simulation.","The lower dominant-term complexity of AMP-A-AC makes it the better choice in regimes where computation time is the bottleneck, while AMP-A-EC is preferable where accuracy matters.","Both algorithms extend approximate message passing to the exact time-domain model of OFDM grant-free access, removing the convergence failure that standard AMP suffers when the pilot length is comparable to or smaller than the number of active devices."],"supporting_citations":[],"fun_headline_variants":["AMP-A-EC vs AMP-A-AC: Accuracy vs speed trade-off","AMP duo: Detect devices and channels in grant-free OFDM","AMP solves convergence for wideband grant-free access"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The methods rely on the factor graph correctly matching the true statistics of time-domain channels and device activities; if the real channel or activity distribution differs from the modeled one, the derived MAP and MMSE objectives are mis-specified and the claimed performance may degrade.","fun_headline_variants_meta":{"raw":{"variants":["AMP-A-EC vs AMP-A-AC: Accuracy vs speed trade-off","AMP duo: Detect devices and channels in grant-free OFDM","AMP solves convergence for wideband grant-free access"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000746,"raw_usage":{"total_tokens":3328,"prompt_tokens":951,"completion_tokens":2377,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":567,"completion_tokens_details":{"reasoning_tokens":2321}},"tokens_in":567,"tokens_out":2377,"duration_ms":21182,"temperature":1.0,"reasoning_tokens":2321,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T04:33:02.570040+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Simulate grant-free access with channel taps drawn from a distribution that differs from the prior assumed in the factor graph (for example, correlated taps or non-Gaussian fading), then compare AMP-A-EC's empirical detection error probability and MSE with its state-evolution prediction. A systematic gap that grows with the mismatch would show the state-evolution analysis and the algorithm's optimality claim depend on the exact prior.","supporting_citations":[],"review_version":1}