{"id":"a4d59c36-d221-4df9-864c-6578a28588f2","arxiv_id":"2506.21664","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":4.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"Under finite blocklength constraints, a RIS-assisted cell-free MIMO system recovers from blockages only beyond critical blocklength thresholds, which shrink as the number of RIS elements grows.","lead":"This paper studies how finite blocklength constraints affect the ability of a RIS-assisted wireless network to recover from blockages. It finds blocklength thresholds beyond which recovery succeeds, and shows that larger RIS arrays lower those thresholds.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"One-iteration alternating SCA may produce the sharp thresholds as a solver artifact rather than a system property; the claimed feasibility boundary is not established.","rationale":"The reader's stated weakest assumption is the unvalidated finite-blocklength normal approximation in Eq. (6). That is a legitimate concern, but it is not the most load-bearing one: at blocklengths of 800-900 symbols the normal approximation is typically accurate to within a few percent, so it is unlikely to destroy the qualitative threshold phenomenon. The deeper problem is that the thresholds are produced by a nonconvex alternating heuristic with one SCA iteration per subproblem and a penalty relaxation for unit-modulus constraints. For the central claim to hold, it must be true that below the threshold no feasible (w,v) exists and above it one does; the paper does not verify feasibility of the returned solution or global optimality of the feasibility boundary. The missing BLER parameter ε further weakens reproducibility of the exact quoted numbers. These concerns do not demand rejection, but they sharpen the conditions under which the paper's headline claim can be accepted: the sharp transition should be shown to persist under randomized restarts and an exact feasibility check. The verdict therefore remains CONDITIONAL, with the condition focused on algorithmic robustness rather than solely on the normal approximation.","tokens_in":9073,"tokens_out":4928,"duration_ms":66224,"concrete_test":"For a fixed channel realization and the exact parameters of Figure 2, at blocklengths η ∈ {800, 830, 838, 850, 900, 907, 920}, run Algorithm 1 from 20 random initializations and also run full alternating SCA to convergence (or verify feasibility via a convex relaxation of (P1) without the unit-modulus constraint). If any blocklength below 838 yields a feasible solution from any initialization, or if the transition point shifts by more than ±5%, the threshold is a solver artifact rather than a system property.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim treats the sharp blocklength thresholds in Figs. 2-3 as a property of the wireless system. The numerical evidence, however, comes from Algorithm 1, which solves the nonconvex problem (P1) by alternating one SCA iteration of (P2) and one of (P3), with no convergence certificate and with the unit-modulus constraint handled by a penalty term Φ. A feasible point for the original problem is never verified: the output (ŵz, v̂z) could violate the unit-modulus or SINR constraints slightly, and the algorithm can stagnate in a local region. Therefore the abrupt jump from 'unrecovered' to 'recovered' at η≈838/907 may be the point at which this particular heuristic finally crosses into the feasible set (or the penalty landscape changes), rather than a fundamental feasibility threshold. The conclusion's 'one-symbol increase can determine success or failure' is especially under-supported, since feasibility of continuous variables under a continuous parameter is generically an open condition; producing a one-symbol jump requires exact global optimization, not a single-pass alternating heuristic. In addition, Section V never specifies the BLER ε in Eq. (6), so even the quoted threshold values are not reproducible from the stated parameters.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies resilience in a cell-free MIMO downlink system with a RIS, where the recovery phase operates under finite-blocklength (FBL) constraints. A resilience metric is formed from absorption, adaptation, and time-to-recovery, and the network adaptation gap is minimized subject to FBL rate constraints and unit-modulus RIS constraints. The optimization is solved by alternating successive convex approximation, with one iteration per subproblem. Numerical results in Section V show sharp transitions in the resilience metric as a function of blocklength, and the paper interprets these as critical blocklength thresholds that separate failed from successful recovery, with RIS size shifting the thresholds. The conclusion states that even a one-symbol blocklength increase can determine recovery success or failure.","tokens_in":9286,"tokens_out":2927,"duration_ms":35006,"significance":"If the threshold phenomenon is real, it is of practical interest for 6G resilience design: it would imply that short-packet recovery has a sharp feasibility boundary and that RIS size can tune that boundary. The paper's formulation is standard and the FBL rate expression is used in a principled way. The numerical exploration directly compares recovery with no adaptation, which is a useful design perspective. The main contribution is the observation of blocklength thresholds and their dependence on RIS size, which would be a meaningful systems insight if supported by reliable numerical evidence.","major_comments":[{"comment":"The central threshold claim in Figures 2 and 3 rests entirely on Algorithm 1, which performs only one SCA iteration per alternating subproblem and handles the unit-modulus constraint through the penalty term Phi in Section IV.B. The paper provides no convergence proof for this scheme and never verifies that the output (w_hat, v_hat) is feasible for the original problem (P1), including constraints (13) and (14). Therefore the sharp jump in the resilience metric from 'unrecovered' to 'recovered' may be an artifact of the heuristic finally crossing into a feasible region, or of a change in the penalty landscape, rather than a system-level feasibility threshold. To support the claim, the authors should either provide a feasibility check for the reported operating points, compare the one-iteration results against fully converged solutions for a subset of blocklengths, or prove that the alternating SCA converges to a feasible stationary point.","section":"Section IV-C and Algorithm 1"},{"comment":"The finite-blocklength achievable rate in Eq. (6) uses the normal approximation with the Gaussian-signaling dispersion in Eq. (7), but the paper gives no error bound or validation of this approximation at the blocklengths used, which range from about 60 to 3000 symbols. At short blocklengths the normal approximation can be inaccurate, so the threshold locations (838 and 907 symbols in Figure 2, and the corresponding values in Figure 3) are not guaranteed. In addition, Section V never specifies the BLER epsilon appearing in Eq. (6), so the numerical results are not reproducible from the stated parameters. Please state epsilon and provide either a comparison with exact FBL bounds or a discussion of the approximation's validity in the simulated blocklength range.","section":"Equations (6)-(7) and Section V"},{"comment":"The concluding statement that 'even a one-symbol increase in blocklength can determine the success or failure of recovery' is not supported by the evidence. The feasibility region of continuous variables under continuous parameters is generically an open condition, so a sharp one-symbol transition would require exact or global optimization, not a single-pass alternating heuristic. The numerical results show a steep but not necessarily discontinuous increase, and the algorithm's limited convergence makes the one-symbol sensitivity claim disproportionate. The authors should either temper the conclusion to describe a steep transition observed with the proposed heuristic, or provide a rigorous argument (e.g., a monotonicity or exactness result) for a true threshold.","section":"Section VI, Conclusion"}],"minor_comments":[{"comment":"The definition of the Q-function has a sign error: the integrand should be exp(-t^2/2), not exp(t^2/2).","section":"Footnote 1"},{"comment":"The Taylor expansion of sqrt(V(q_k)) appears to have a misplaced parenthesis or a missing factor in the second term; please double-check the derivative expression against the standard derivative of sqrt(1 - (1+q)^-2).","section":"Equation (20)"},{"comment":"The pseudocode is difficult to read because the control flow arrows and assignments are garbled; please rewrite it as a structured algorithm with clear loops and termination conditions.","section":"Algorithm 1"},{"comment":"The simulation setup does not specify Tcalc, the coherence time Tc, or the BLER epsilon, all of which are needed to reproduce the results and to understand the algorithm's stopping behavior.","section":"Section V"},{"comment":"The word 'adaption' is used repeatedly; it should be 'adaptation' for consistency with the rest of the text.","section":"Throughout"},{"comment":"The paper relies heavily on the authors' earlier works for the alternating SCA framework; please clarify explicitly what is new in the present formulation beyond the FBL rate expression in the adaptation metric, so that the incremental contribution is clear.","section":"Section IV-C and References [6, 11]"}],"recommendation":"major_revision","confidential_remarks":"The paper addresses a timely and relevant topic, and the threshold observation is interesting. However, the main numerical claim is not yet backed by sufficient algorithmic guarantees or validation. The missing BLER parameter and the lack of a feasibility check are concrete fixable issues. I recommend major revision rather than rejection because the central idea is plausible and the concerns can be addressed within the manuscript's scope."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things you should know. The paper's central observation—a sharp blocklength threshold in their resilience metric, shifting with RIS size—is new and clearly presented. The evidence for it, however, is not as solid as the text suggests, because the thresholds come from a one-iteration alternating SCA heuristic with no feasibility check, and the BLER parameter is never specified.\n\nThe genuine strength is the question: when should a system attempt FBL-based recovery instead of absorbing a disruption? The authors take the standard FBL rate expression from Polyanskiy et al., embed it in a resilience framework they have been developing, and show in simulation that there is a narrow blocklength region where recovery abruptly becomes successful. The system model is standard, the optimization is a reasonable heuristic for a nonconvex problem, and the figures tell a coherent story. The claim that larger RIS arrays lower the threshold blocklength is intuitive and consistent with the model. The writing is clear, and the citation pattern is appropriate—most self-citations are to the framework being extended.\n\nThe soft spots are real. First, the stress-test concern lands: Algorithm 1 alternates single SCA iterations, handles unit-modulus constraints through a penalty, and never verifies that the output satisfies the original constraints. The sharp jump may be where this particular heuristic crosses into the feasible set, not a fundamental system threshold. The conclusion's 'one-symbol increase can determine success or failure' is especially fragile, since feasibility in continuous variables is generically open; a one-symbol cliff would require the algorithm to behave discontinuously. The authors should check feasibility at each point or solve the problem exactly for a few blocklengths to support the claim. Second, the FBL normal approximation is used down to blocklengths near 60 without validation, and the BLER ε in Eq. (6) is missing, so the threshold values (838, 907) cannot be reproduced from the paper. Third, the plots show no error bars or averaging over channel realizations, which makes the sharp transitions hard to assess.\n\nIf the threshold phenomenon survives scrutiny, it is a useful design criterion for 6G resilience. As it stands, this is a plausible simulation study with an under-validated central claim. I would not cite it for the specific numbers, but it deserves a serious referee: the question is timely, the framework is clear, and the gaps are fixable. Send it to review, asking for feasibility checks, error bars, and the BLER value.","headline":"A crisp simulation study on FBL recovery thresholds, but the sharp transitions are not yet shown to be system properties rather than solver artifacts; worth a careful revision.","tokens_in":9839,"tokens_out":4353,"would_cite":false,"duration_ms":48793,"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":"Blocklength thresholds decide when wireless recovery is worth attempting.","keywords":["finite blocklength","wireless resilience","reconfigurable intelligent surface","cell-free MIMO","blocklength threshold","short-packet communication","adaptation gap","ultra-reliable low-latency communication"],"falsifier":"Simulate or evaluate non-asymptotic achievability and converse bounds at the SINR values attained just below and above the claimed thresholds, blocklengths $\\eta = 838$ and $\\eta = 907$ with $r^{\\mathrm{des}}_k = 35$ and $37$ Mbps, and compare the actual achievable rates against Eq. (6); if the true transition blocklength differs substantially or no transition exists, the threshold claim fails.","tokens_in":8817,"feed_emoji":"📡","tokens_out":7133,"duration_ms":75399,"temperature":0.7,"pith_summary":"This paper argues that the decision to attempt fast recovery from a wireless outage is blocklength-driven: below a critical number of channel uses, switching to short packets for recovery loses more rate to finite-blocklength coding than the outage itself costs, so the system should absorb the disruption instead; above that threshold, recovery pays off and the system becomes more resilient overall. The setting is a cell-free MIMO downlink assisted by a reconfigurable intelligent surface (RIS), and the recovery phase is modeled with the finite-blocklength (FBL) rate formula, in which short codewords pay a rate penalty that shrinks as the blocklength grows. In the simulated scenarios, the threshold appears at about 838 symbols for a 35 Mbps quality-of-service target and at about 907 symbols for 37 Mbps, while a 40 Mbps target cannot recover even at 3000 symbols. Increasing the number of RIS reflecting elements shifts these thresholds to shorter blocklengths, and increasing the target rate shifts them to longer ones. The paper presents this as a design principle: accounting for blocklength in the resilience metric is what lets a network know whether recovery will help or hurt.","feed_headline":"Blocklength cutoff decides whether wireless recovery succeeds","feed_subtitle":"At 838–907 symbols recovery flips from harmful to helpful; more RIS elements lower the cutoff.","key_machinery":"The load-bearing object is the finite-blocklength achievable-rate constraint in Eq. (6), the normal approximation $r_k \\le B(\\log_2(1+\\Gamma_k) - \\Omega \\sqrt{V(\\Gamma_k)/\\eta})$ with channel dispersion $V(\\Gamma_k)=1-(1+\\Gamma_k)^{-2}$, embedded in an optimization that minimizes the adaptation gap $\\Psi=\\sum_k |r_k/r^{\\mathrm{des}}_k - 1|$ over beamformers and RIS phase shifts. The resilience metric $r=\\lambda_1 r_{\\mathrm{abs}}+\\lambda_2 r_{\\mathrm{ada}}+\\lambda_3 r_{\\mathrm{rec}}$ then compares absorbing the outage against adapting with short packets. The nonconvex problem is solved by alternating optimization with one-step successive convex approximation, so the solution can be recomputed within a coherence time; the FBL rate expression is what couples blocklength, SINR, target rate, and recovery time and produces the sharp feasibility transition.","core_discovery":"On the paper's own terms, the central discovery is that finite blocklengths turn resilience from a smooth trade-off into a threshold phenomenon. For a given quality-of-service demand, there exists a critical blocklength $\\eta$ at which the resilience metric $r$ jumps from the no-adaptation level to a recovered level, because only then can the reconfiguration compensate both the rate penalty from Eq. (6), $r_k \\le B(\\log_2(1+\\Gamma_k)-\\Omega\\sqrt{V(\\Gamma_k)/\\eta})$, and the loss from the blocked link. Numerical evaluation shows this threshold at $\\eta\\approx 838$ symbols for $r^{\\mathrm{des}}_k = 35$ Mbps and $\\eta\\approx 907$ symbols for $r^{\\mathrm{des}}_k = 37$ Mbps with $M=1000$ RIS elements; at $40$ Mbps the system stays unrecovered up to $\\eta=3000$. With the target held at $37$ Mbps, an $M=625$ RIS fails to recover, while $M=1000$ and $M=1600$ both cross the threshold, with the larger array crossing at a shorter blocklength. The paper therefore claims that resource surplus and RIS size are not just performance boosters; they are what determine whether fast recovery is feasible at all.","pith_inferences":["The threshold is probably not special to RIS: any mechanism that raises the achievable SINR, such as better beamforming, more antennas, or interference cancellation, should shift the same threshold to shorter blocklengths, so the paper's threshold vocabulary could be used to compare physical-layer resilience technologies.","If the normal approximation in Eq. (6) is replaced by non-asymptotic bounds, the thresholds may move numerically, but the qualitative bistable structure, where recovery either fully fails or fully succeeds at a critical blocklength, would likely survive because it comes from the rate constraint becoming feasible.","A practical controller could exploit the sharpness by probing at a few blocklengths near the suspected threshold to estimate whether the network has enough resource headroom to recover, without running the full optimization.","The one-symbol sensitivity suggests that specifying blocklength to the symbol, rather than to the nearest frame, should be part of recovery signaling in standards-oriented designs."],"forward_implications":["Operators can treat the existence of a threshold as a decision rule: below it, initiating recovery with short packets is predicted to lower the resilience metric, so absorbing the outage is the better action.","The threshold position depends on the QoS target, so tuning the desired rate downward is a concrete way to make recovery feasible at shorter blocklengths.","Larger RIS deployments shift the threshold to shorter blocklengths, meaning physical-layer reconfiguration effort and temporal responsiveness are traded against each other.","At the transition, a one-symbol increase in blocklength can move the system from no recovery to full recovery, so the fine granularity of blocklength selection matters in resilience-critical operation.","The same framework yields a comparison baseline: when no blocklength up to 3000 produces recovery, as in the 40 Mbps case, the model says the network should deliberately ignore the disruption."],"supporting_citations":[{"why":"Supplies the finite-blocklength rate formula in Eq. (6), whose rate penalty creates the thresholds.","marker":"[8]"},{"why":"Defines the three resilience sub-metrics combined in Eq. (11).","marker":"[4]"},{"why":"Provides the one-iteration alternating-SCA recovery procedure used in Algorithm 1.","marker":"[6]"},{"why":"Contributes the adaptation-gap objective, the SCA treatment of SINR constraints, and the penalty method for unit-modulus RIS phases.","marker":"[11]"},{"why":"Provides the Taylor approximation of the dispersion square root used to convexify constraint (17).","marker":"[16]"},{"why":"Supplies the correlated RIS channel model used in the simulations behind Figures 2 and 3.","marker":"[17]"}],"fun_headline_variants":["Blocklength cliff flips wireless recovery from harmful to helpful","RIS elements push recovery's symbol threshold lower","Symbol cutoff decides if wireless recovery backfires","Wireless recovery only works past a finite symbol cliff"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The analysis assumes the finite-blocklength normal approximation in Eq. (6), which has no proven error bound at the short blocklengths used here, is accurate enough to locate the thresholds; if it is off at blocklengths around 60 to 900 symbols, the claimed threshold positions, and possibly their existence, could change.","fun_headline_variants_meta":{"raw":{"variants":["Blocklength cliff flips wireless recovery from harmful to helpful","RIS elements push recovery's symbol threshold lower","Symbol cutoff decides if wireless recovery backfires","Wireless recovery only works past a finite symbol cliff"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001001,"raw_usage":{"total_tokens":4288,"prompt_tokens":1053,"completion_tokens":3235,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":669,"completion_tokens_details":{"reasoning_tokens":3175}},"tokens_in":669,"tokens_out":3235,"duration_ms":24858,"temperature":1.0,"reasoning_tokens":3175,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T22:21:50.237023+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Simulate or evaluate non-asymptotic achievability and converse bounds at the SINR values attained just below and above the claimed thresholds, blocklengths $\\eta = 838$ and $\\eta = 907$ with $r^{\\mathrm{des}}_k = 35$ and $37$ Mbps, and compare the actual achievable rates against Eq. (6); if the true transition blocklength differs substantially or no transition exists, the threshold claim fails.","supporting_citations":[{"cited_title":"Channel coding rate in the finite blocklength regime,","cited_arxiv_id":null,"evidence_quote":"Supplies the finite-blocklength rate formula in Eq. (6), whose rate penalty creates the thresholds."},{"cited_title":"Accelerated Recovery with RIS: Designing Wireless Resilience in Mission-Critical Environments","cited_arxiv_id":"2504.11589","evidence_quote":"Provides the one-iteration alternating-SCA recovery procedure used in Algorithm 1."},{"cited_title":"Max-min fairness and PHY-layer design of uplink MIMO rate-splitting multiple access with finite blocklength,","cited_arxiv_id":null,"evidence_quote":"Provides the Taylor approximation of the dispersion square root used to convexify constraint (17)."},{"cited_title":"Rayleigh fading modeling and channel hardening for reconfigurable intelligent surfaces,","cited_arxiv_id":null,"evidence_quote":"Supplies the correlated RIS channel model used in the simulations behind Figures 2 and 3."}],"review_version":1}