{"id":"f0f20f9a-b85d-44af-9793-d774a0c0b25a","arxiv_id":"2607.13710","paper_version":1,"verdict":"REJECT","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"high","formal_verification":"none","parameter_count":6,"one_line_summary":"A virtual-queue resilience metric and a two-phase beamforming/RIS/blocklength optimizer that make transmission time a resource for recovering from wireless disruptions.","lead":"This paper proposes a resilience framework for wireless networks that treats transmission duration as a controllable resource in short-packet (finite-blocklength) links, and introduces a metric that scores absorption, adaptation, and recovery after disruptions. It gives 6G/URLLC designers a common yardstick for comparing how different physical-layer mechanisms handle outages and traffic surges.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Blocklength updates (51), (58)-(60) minimize by stationary points of concave functions; these stationary points are maxima, so the central dynamic-blocklength mechanism is not solving the stated optimization.","rationale":"The reader's weakest assumption is exactly the load-bearing issue. The manuscript's central novelty is dynamic blocklength adaptation as a recovery mechanism, and the closed-form updates (51), (58), (59), (60) are the analytical core of that mechanism. If those updates select stationary points of concave functions, they maximize the very quantities the problems claim to minimize, or at best select an interior point with no optimality justification. The same sign/structure error appears in the adaptation-phase stability constraint, where the small-root branch of the quadratic inequality is not the branch that corresponds to stable high-rate transmission. The additional mismatch in Alg. 1 — using adaptation formulas during the absorption phase — aggravates the inconsistency. Because the numerical results in Fig. 4 are generated by an algorithm whose stated blocklength update does not solve the stated problem, the empirical resilience comparisons cannot validate the proposed approach without a corrected derivation and preferably reproducible code. I therefore see no reason to alter the reader's REJECT verdict; the central optimization claim is not supported as submitted.","tokens_in":24522,"tokens_out":11767,"duration_ms":131779,"concrete_test":"Independently re-derive (51) from (50) by checking second-order conditions: compute d²δbar_k/dη² from (65) for the Sec. V parameters after the first outage. Because it is negative for C_k>α_des, the stationary point is a maximum; then evaluate the single-user objective (49) at ηmin=128, ηmax=512, and η* from (51) to confirm the endpoint is lower. For the adaptation phase, re-derive (58) by checking that d²f/dx² = 2Σ Q_k(α_des-C_k) is negative in the stable regime, so x* is a maximizer. A corrected derivation must replace (51) and (58) by an explicit comparison over the interval endpoints.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Sec. IV-B derives (51) as the solution of (P3): min_η max_k δbar_k. From (65), δbar_k(η)=(-A_k η+B_k√η)/D_k with A_k=C_k-α_des, B_k=BΩ√V, D_k>0. For A_k>0, d²δbar_k/dη² = -B_k/(4D_k η^{3/2}) < 0, so δbar_k is concave in η (and in √η). The first-order point x*=B_k/(2A_k) is therefore the maximizer of the active user's violation, not the minimizer; the minimum of a concave function over [ηmin,ηmax] is at an endpoint. For A_k<0 there is no positive stationary point, and (51)'s squaring manufactures a spurious interior solution. The adaptation objective in (70)/(71) has the same concave-in-√η structure when C_k>α_des, so (58)/(73) maximizes drift rather than minimizing it. The stability inequality (75) is a convex quadratic in x=√η; its ≥0 feasible set is x≤small-root ∪ x≥large-root, and the branch relevant to stable high-rate operation is the large-root branch, not (76). Moreover, Alg. 1 lines 10-11 use the adaptation formulas (58)-(60) during the absorption phase even though Sec. IV-B derived (51) for absorption. Since dynamic blocklength is the proposed resilience mechanism, these errors undermine Fig. 4(d) and the comparative resilience claims.","agreement_with_reader":"agree"},"referee_report":null,"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: the resilience metric and the two-phase absorption/adaptation framing are real contributions, but the closed-form blocklength updates that drive the whole mechanism are mathematically wrong, and the algorithm contradicts its own derivation. I side with the reader's rejection.\n\nWhat's good: the virtual-queue view of resilience is clean. Defining absorption loss, adaptation efficiency, and relative recovery index as in (25), (28), (29) and combining them in (30) is genuinely useful — interpretable and easy to compute. The system model is thorough: cell-free MIMO with RIS, imperfect CSI, Poisson traffic, FBL rate expressions, and a careful queueing setup. The numerical section is also comprehensive, covering repeated outages, AI-driven surges, different RIS sizes, and imperfect CSI. If the optimization machinery were sound, this would be a solid contribution to the URLLC/6G literature.\n\nWhere it falls down: the stress-test note is correct on every point. In Appendix VII-A, (65) is concave in √η when A_k > 0, so the stationary point found in (67) is the maximizer of δbar_k, not the minimizer. Minimizing a concave function over an interval happens at a boundary. Thus (51) is not the solution to (P3). Same error in Appendix VII-B: the drift objective (70) is concave in x, so (58)/(73) finds the maximum drift, not the minimum. The stability inequality (75) is a convex quadratic; the feasible set for stability is x ≤ small-root or x ≥ large-root, and the branch that matters for high-rate operation is the large-root branch, not (76). Also, Algorithm 1 lines 10-11 compute η using (58)-(59) during the absorption phase, even though (51) was derived for absorption. That inconsistency means the algorithm's blocklength choice is not the one the paper analyzes in Section IV-B.\n\nBecause dynamic blocklength adaptation is the stated resilience mechanism, these errors are load-bearing. The numerical plots in Fig. 4(d) and the comparative resilience claims cannot be taken as validating the framework. I would not reject the underlying idea — the metric and the two-phase framing deserve a second chance — but the paper as submitted needs a major revision. A corrected version should solve the blocklength subproblem properly (likely by checking endpoints and using the correct root of the quadratic) and align the algorithm with the derivations.\n\nI would still send this to peer review: the ideas are worth referee time, and a good report will point the authors precisely to the fix. I would not cite it in its current form, and I wouldn't hold it up as an example of clean derivations. Set would_accept_peer_review = true.","headline":"Useful metric and framing, but the blocklength updates minimize concave functions at their maxima, so the central mechanism is unsupported.","tokens_in":25431,"tokens_out":5294,"would_cite":false,"duration_ms":55836,"reading_group":"maybe","serious_thinker":"no","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper claims that wireless resilience in URLLC networks can be engineered by treating blocklength as a dynamic control variable, jointly optimized with beamforming and RIS phase shifts.","keywords":["resilience","finite blocklength","URLLC","virtual queues","reconfigurable intelligent surface","cross-layer optimization","Lyapunov drift","blocklength adaptation"],"falsifier":"Recompute the stationary point from Appendix VII-A: with A_k = C_k − α_des, the per-user gap δ̄_k as a function of √η is a downward-opening quadratic, so its derivative zero is a maximum. For the Fig. 4 parameters, evaluate δ̄_k at the formula's value, at η_min, and at η_max; if an endpoint gives a smaller worst-case gap, (51) is not the minimizer. Then rerun Algorithm 1 with an exhaustive search over [128, 512] for η and compare the QSI trajectory against the closed-form version; a material difference would show the reported resilience gains depend on the disputed update.","tokens_in":24357,"feed_emoji":"📡","tokens_out":8226,"duration_ms":232918,"temperature":0.7,"pith_summary":"The paper tries to establish that resilience to wireless disruptions is not only a matter of power or coding but of time: in the finite-blocklength regime, the duration of a transmission is itself a controllable resource. It builds a virtual-queue model that turns a service deficit into an accumulated backlog, and a metric that scores a system on how well it absorbs a disturbance, adapts its resources, and recovers its queues. On top of this, it develops a two-phase alternating-optimization algorithm that adjusts beamforming, RIS phase shifts, and blocklength, first to stabilize the queues and then to drain the accumulated deficit. The central claim is that dynamic blocklength adaptation is what lets the system fully recover from repeated channel outages and AI-driven traffic surges, and that the proposed metric gives a consistent, interpretable way to compare resilience across different strategies and disruption types.","feed_headline":"Blocklength adaptation restores wireless queues after repeated outages","feed_subtitle":"A two-phase optimization turns transmission duration into a controllable recovery resource.","key_machinery":"The load-bearing mechanism is the variable-length time slot: with bandwidth fixed, a codeword of length η takes T_q = η/B seconds, so choosing the blocklength is equivalent to choosing how long a recovery step lasts and how large the finite-blocklength rate penalty is. The framework tracks service deficits in a virtual queue driven by expected arrivals rather than random ones, and uses a log-barrier Lyapunov drift term to sense when a user is approaching overflow. The optimization alternates between beamforming via successive convex approximation, RIS phase-shift updates via a penalty method, and a closed-form blocklength update, with the blocklength update acting as the temporal control tha","core_discovery":"The paper's central claim is that a wireless URLLC system can absorb and recover from abrupt disturbances by explicitly optimizing how long each transmission lasts. Because the achievable rate in the finite-blocklength regime carries a penalty that shrinks as the blocklength grows, the system can use a longer blocklength during a disruption to reduce that penalty and stop queue growth, then switch to a shorter blocklength during recovery to transmit more aggressively and drain the backlog. The paper formalizes this as a two-phase problem — absorption, then adaptation — and claims that jointly optimizing beamforming, RIS phase shifts, and blocklength keeps queues stable through repeated outag","pith_inferences":["The closed-form blocklength formulas (51) and (58)–(60) are obtained by setting the derivative of the stability-gap function to zero, but that function is concave in the square root of the blocklength, so the stationary point is a maximum rather than a minimum; if so, the true blocklength minimizer lies at an endpoint of the allowed range.","Because the metric and the two-phase formulation are separable from the closed-form solver, a clean test is to replace the blocklength formulas in Algorithm 1 with a simple search over [η_min, η_max]; if the reported recovery behavior changes materially, the formulas are carrying the result, and if not, the framework's qualitative claims survive.","The paper's imperfect-CSI results point to an extension it leaves implicit: building channel-estimation error into the rate constraints as a robustness margin would likely change when recovery is possible, and the same metric could quantify the benefit."],"forward_implications":["If the framework holds, a network can survive repeated direct-link outages that a fixed-blocklength policy cannot survive, because the blocklength can be raised during absorption to cut the FBL rate penalty and restore stability.","The two-phase structure implies that resilience requires deliberately over-serving after an outage — pushing rates above the nominal target to drain accumulated backlog — not just returning to nominal service.","The metric's decomposition means channel outages and traffic surges can be scored on the same absorption–adaptation–recovery scale, revealing which phase limits a given system.","The paper's comparisons imply that systems without dynamic blocklength control remain stable at best and permanently degraded at worst, with queues stuck above their desired operating point even when they do not overflow."],"fun_headline_variants":["Tune transmission duration to survive wireless outages","Longer blocklengths absorb disruptions, shorter ones speed recovery","Blocklength as a recovery dial for wireless queues","Finite blocklength: a lever for wireless resilience","Time-aware optimization keeps URLLC queues stable"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The blocklength-update formulas assume that the best blocklength is the point where the stability-gap function stops changing, but that function curves downward, so the point they identify is a maximum of the gap, not the minimum; the true worst-case-minimizing blocklength sits at one of the allowed endpoints.","fun_headline_variants_meta":{"raw":{"variants":["Tune transmission duration to survive wireless outages","Longer blocklengths absorb disruptions, shorter ones speed recovery","Blocklength as a recovery dial for wireless queues","Finite blocklength: a lever for wireless resilience","Time-aware optimization keeps URLLC queues stable"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000448,"raw_usage":{"total_tokens":2094,"prompt_tokens":738,"completion_tokens":1356,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":482,"completion_tokens_details":{"reasoning_tokens":1282}},"tokens_in":482,"tokens_out":1356,"duration_ms":15486,"temperature":1.0,"reasoning_tokens":1282,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-02T03:57:06.190654+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Recompute the stationary point from Appendix VII-A: with A_k = C_k − α_des, the per-user gap δ̄_k as a function of √η is a downward-opening quadratic, so its derivative zero is a maximum. For the Fig. 4 parameters, evaluate δ̄_k at the formula's value, at η_min, and at η_max; if an endpoint gives a smaller worst-case gap, (51) is not the minimizer. Then rerun Algorithm 1 with an exhaustive search over [128, 512] for η and compare the QSI trajectory against the closed-form version; a material difference would show the reported resilience gains depend on the disputed update.","supporting_citations":[],"review_version":1}