{"id":"9919363d-480d-4c67-a20e-9e855337a53c","arxiv_id":"1908.08872","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"Under a Normal approximation to the SNR distribution, session resource request probabilities in 3GPP mmWave New Radio can be expressed through error functions, and shadow fading strongly shifts the request distribution.","lead":"The paper shows that the signal-to-noise ratio in a 3GPP millimeter-wave cell can be approximated by a Normal distribution, which turns the resource-request distribution into simple error functions. This gives network modelers a fast closed-form tool for estimating how many radio resource blocks a 5G session needs.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Eq. (16) mixes blocked and non-blocked SNR CDFs with a distance-averaged blockage probability, ignoring the correlation between blockage and distance; the 'exact' benchmark used to validate the Normal approximation may therefore be wrong.","rationale":"The reader's weakest-assumption analysis correctly identifies the core issue: Eq. (16) uses a single distance-averaged blockage probability to combine blocked and non-blocked SNR CDFs, even though the blockage probability in Eq. (14) increases with distance. This creates a correlation between the link state and the SNR level that the paper's mixture ignores. Since this approximate mixture is the ground truth against which the Normal approximation is validated, the reported K-S statistics and Table 4 accuracy numbers are not yet evidence about the true model. That is a substantive, load-bearing concern rather than a stylistic one. I do not see a more fundamental flaw: the paper's idea of approximating the SNR CDF by a Normal distribution with mean E[S_dB] and variance sigma_SF^2 + sigma_SdB^2 is plausible, and the error-function resource-request pmf is a useful computational shortcut if the approximation holds. The authors also provide a parameter-light model and concrete numerical comparisons, which is positive. The fix is well-defined: recompute the exact distance-conditioned mixture and compare against both Eq. (16) and the Normal approximation. If the comparison holds, the central claim survives; if not, the accuracy figures need to be revised. This matches the reader's CONDITIONAL verdict, so no change to the verdict is needed.","tokens_in":10048,"tokens_out":3441,"duration_ms":35840,"concrete_test":"Recompute the reference SNR CDF with the distance-dependent blockage probability placed inside the integral: F_S(s) = ∫_0^{rA} [pB(x) F_{S|B,x}(s) + (1 - pB(x)) F_{S|nB,x}(s)] f_X(x) dx, using Eq. (14) for pB(x) and the same parameters as Table 3 for pC = 0.01, 0.05, 0.1. Then compute the Kolmogorov-Smirnov statistic between this corrected benchmark and (a) Eq. (16) and (b) the Normal mixture approximation used in Section 4. If the Normal approximation still matches the corrected benchmark to within K-S ≈ 0.03, and the mean resource-request values in Table 4 shift by less than about 2%, the concern is resolved. If Eq. (16) itself differs from the corrected benchmark by more than the reported K-S value, the validation is measuring the wrong target and the paper's accuracy claims need revision.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's validation of the Normal approximation rests on comparing it to the SNR CDF constructed in Eq. (16). That CDF is WS(x) = pB WSB(x) + (1 - pB) WSnB(x), where pB is the distance-averaged blockage probability from Eq. (15) and WSB, WSnB are the distance-unconditioned CDFs for blocked and non-blocked links. But Eq. (14) gives blockage probability pB(x) that increases with 2D distance x, while longer distances also produce lower SNR. The correct unconditional CDF is F_S(s) = ∫ [pB(x) F_{S|B,x}(s) + (1 - pB(x)) F_{S|nB,x}(s)] f_X(x) dx. Eq. (16) instead evaluates F = pB F_B + (1-pB) F_nB, where F_B and F_nB already average over x. This is exact only if pB(x) is independent of x, which it is not. Because blocked states are more probable at larger x, the blocked SNR component in the true mixture is shifted toward lower SNR relative to the unconditioned F_B used in Eq. (16). The reported K-S statistics (0.01-0.03) and the close mean resource requests in Table 4 are computed against this Eq. (16) benchmark, not against the actual distance-dependent mixture. Thus the central claim that the Normal approximation accurately captures the true SNR CDF is currently supported only by comparison to an approximate reference, and the magnitude of the resulting error is unreported. This is a load-bearing gap in the validation of the main result, though it is addressable by recomputing the benchmark.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops an analytical approximation to the session resource-request probability mass function (pmf) in a single-sector millimeter-wave 3GPP New Radio deployment. The model assumes uniformly distributed user equipment, 3GPP-like distance-dependent path loss with two link states (blocked/non-blocked), dynamic blockage, and lognormal shadow fading. The authors derive closed-form SNR CDFs for the blocked and non-blocked states, mix them using a distance-averaged blockage probability in Eq. (16), and then approximate the resulting CDF by a mixture of Normal distributions whose mean and standard deviation are moment-matched to the analytical model. This lets the resource-request pmf be evaluated through error functions. Numerical results report Kolmogorov-Smirnov statistics between 0.01 and 0.03 for the Normal fit and mean resource requests within a few percent in Table 4, and the paper highlights that omitting shadow fading changes the pmf substantially.","tokens_in":10325,"tokens_out":7850,"duration_ms":82165,"significance":"If the central claim is correct, the paper provides a useful computational shortcut: replacing a numerical convolution over user location and shadow fading with a closed-form Normal-mixture approximation would speed up queueing and stochastic-geometry analyses of 5G NR sessions. The derivation of the closed-form SNR CDF in Eq. (13) is a genuine technical contribution, and the explicit moment matching makes the approximation transparent. However, the validation benchmark in Eq. (16) is itself questionable because it mixes distance-unconditioned CDFs with a distance-averaged blockage probability even though the blockage probability in Eq. (14) is distance-dependent. The reported K-S statistics and Table 4 errors are therefore measured against a possibly biased reference; the magnitude of this bias is unreported. The issue is addressable, but it is load-bearing for the paper's main claim, so the manuscript requires major revision.","major_comments":[{"comment":"The mixture W_S(x) = p_B W_SB(x) + (1-p_B) W_SnB(x) is not the correct unconditional SNR CDF when the blockage probability depends on distance. Equation (14) gives p_B(x) increasing with 2D distance x, while the path loss in Eq. (2) also depends on distance. The correct unconditional CDF is F_S(s) = ∫ [p_B(x) F_{S|B,x}(s) + (1-p_B(x)) F_{S|nB,x}(s)] f_X(x) dx, whereas Eq. (16) computes p_B ∫ F_{S|B,x}(s) f_X(x) dx + (1-p_B) ∫ F_{S|nB,x}(s) f_X(x) dx. The difference is Cov(p_B(X), F_{S|B,X}(s) - F_{S|nB,X}(s)), which is generally non-zero because both p_B(x) and the conditional SNR CDFs vary with x. Since Eq. (16) is the 'exact' benchmark used in Section 4 to assess the Normal approximation, the reported K-S values and Table 4 errors may not reflect the true accuracy of the approximation. Please replace the benchmark with the correct distance-conditioned mixture, or validate against Monte Carlo simulation, and recompute the reported statistics.","section":"Section 3.2, Eq. (16)"},{"comment":"The Normal approximation is parameterized by the first two moments of the analytical model and then compared with that same analytical model. Agreement in mean and standard deviation is therefore partly by construction; the K-S statistic is informative about shape beyond the first two moments, but it is an in-sample metric. More importantly, the comparison is made against Eq. (16), which, per the comment above, is itself an approximation. Please report the distance-conditioned benchmark (or a simulation-based benchmark) and the resulting K-S statistics and pmf errors, including the maximum absolute error or total variation distance between the approximate and correct resource-request pmfs.","section":"Section 4.1, Fig. 2 and Table 4"}],"minor_comments":[{"comment":"The variable d_E is used in Eq. (9) but is not defined; it appears to denote the coverage radius r_A. Please define it explicitly.","section":"Section 2, Eq. (9)"},{"comment":"The symbol gamma appears in the argument of the error functions in Eq. (13) but is not defined; it should be the path-loss exponent zeta introduced in Eq. (5).","section":"Section 3.2, Eq. (13)"},{"comment":"The row labeled 'Noise figure, WP RB 1.44 Mhz' mixes the noise figure with the PRB bandwidth and has incorrect units; it should read 'PRB bandwidth, 1.44 MHz' as a separate parameter.","section":"Table 3"},{"comment":"The K-S values for the without-shadow-fading case (0.13-0.21) are not discussed in the text; a sentence explaining why the Normal fit degrades in that case would help the reader interpret the role of shadow fading.","section":"Section 4.1, Fig. 2"},{"comment":"The conclusion refers to the session resource requirements 'pdf', but the object is a probability mass function (pmf) because the number of requested resources is discrete.","section":"Section 5"}],"recommendation":"major_revision","confidential_remarks":"The core issue is not stylistic: Eq. (16) is the reference distribution used to validate the paper's main claim, and the distance-averaged blockage mixing is only exact if blockage is independent of distance, which Eq. (14) contradicts. The fix is straightforward in principle (condition on distance before mixing, or simulate), and if the corrected benchmark still yields K-S statistics around 0.01-0.03 and pmf errors of a few percent, the paper would be a solid contribution. I would encourage the editor to request a revision focusing on this validation gap rather than reject on the basis of it."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear colleague,\n\nThe one thing to know: this paper offers a genuinely handy closed-form approximation for the SNR CDF in 3GPP mmWave NR with shadow fading, and converts it into a session resource-request pmf via error functions. That is a useful tool for people doing queueing or stochastic geometry on 5G NR. The soft spot is the validation benchmark: Eq. (16) mixes blocked and non-blocked CDFs with a distance-averaged blockage probability even though Eq. (14) says blockage depends on distance. That needs repair before I'd fully trust the headline numbers.\n\nWhat is new and good: the closed-form SNR CDF with lognormal shadowing (Eq. 13) and the observation that the resulting CDF is well approximated by a Normal, so the resource pmf becomes a sum of error functions. The reported K-S statistics (0.01\\u20130.03) genuinely show the Normal shape is close, and Table 4's mean resource requests are within a couple percent. For modelers needing a fast pmf, this is valuable.\n\nThe soft spots, in order of seriousness:\n\n1. The Eq. (16) issue is real. The correct unconditional CDF should condition the distance distribution on the link state before mixing, because blocked states are more likely at larger distances. The stress-test note is right. The paper's \"exact\" CDF is itself an approximation, and the reported accuracy is measured against that approximate benchmark. The magnitude of the error is unreported. This is addressable but load-bearing.\n\n2. All validation is against the paper's own analytical model. No simulation, no measurement. The Normal parameters are moment-matched to that same model, so mean agreement is partly by construction. The K-S values do assess shape, but an independent check would be far more convincing.\n\n3. Notation and typos hurt reproducibility: zeta is never given a numeric value (though implied by the 21 in Eq. 2), Eq. (5)'s distance term looks like it is missing a square, and Eq. (13) is unwieldy. These are minor but worth cleaning.\n\n4. The paper does not position itself against prior normal/lognormal approximation work. It would be stronger if it explained what is different here.\n\nFor the right reader\\u2014someone building an analytical mmWave NR model\\u2014this is a useful building block. I would send it to peer review, but require the reference CDF to be corrected or justified and ask for a simulation sanity check. As it stands, the central approximation may well hold, but the evidence does not yet fully support it.\n\nBest,\n[Your name]","headline":"Useful closed-form approximation for mmWave NR resource pmf, but the validation benchmark mixes blockage with a distance-averaged probability and that needs fixing.","tokens_in":10941,"tokens_out":3778,"would_cite":false,"duration_ms":40752,"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":"The paper claims that SNR at a random user in 3GPP millimeter-wave New Radio closely follows a Normal mixture, making session resource-request distributions computable with error functions.","keywords":["5G New Radio","millimeter-wave communications","SNR distribution approximation","Normal distribution mixture","shadow fading","resource request probability mass function","blockage modeling","error function approximation"],"falsifier":"Simulate a millimeter-wave sector with explicit blockers whose positions correlate with user distance, compute the empirical SNR CDF conditioned on blocked and unblocked states, and compare it with the paper's Eq. (16) mixture; if the Kolmogorov-Smirnov distance rises well above the reported 0.01 to 0.03, the independence assumption is the reason.","tokens_in":9759,"feed_emoji":"📶","tokens_out":6229,"duration_ms":57562,"temperature":0.7,"pith_summary":"This paper tries to establish a fast analytical substitute for the session resource-request distribution in millimeter-wave 5G New Radio systems. It claims that the cumulative distribution function of the signal-to-noise ratio at a randomly located user, under the 3GPP propagation model with shadow fading and moving blockers, is closely approximated by a mixture of Normal distributions. That matters because the resource-request probability mass function, needed to size base-station resources in queuing models, cannot be written with elementary functions otherwise. Once the SNR CDF is Normal, the resource-request pmf follows from the modulation-and-coding mapping through error functions, and the reported Kolmogorov-Smirnov distances for the SNR fit are between 0.01 and 0.03.","feed_headline":"A Normal curve fits 5G mmWave SNR almost exactly","feed_subtitle":"Closed-form error functions then give the session resource-request distribution without numerical integration.","key_machinery":"The load-bearing construction is the mixture CDF $W_S(x)=p_B W_{S,B}(x)+(1-p_B)W_{S,\\mathrm{nB}}(x)$, where each component CDF comes from Eq. (13): the distance-induced SNR distribution of a uniformly located UE is transformed into a density via the random-variable method, then convolved with a zero-mean Normal shadow-fading term. The weight $p_B$ is the distance-averaged blockage probability from the random-direction-mobility blocker model. This construction matters because the convolution with Gaussian shadow fading yields an expression in error functions, and the final mixture is then replaced by Normal components with matched first two moments, giving a closed-form route to $m_j$ and hence to the resource-request pmf.","core_discovery":"The central discovery is that Lognormal shadow fading smooths the distance-induced SNR distribution into a nearly Gaussian shape. In decibels, the SNR at a uniformly located UE is represented as $S_{\\mathrm{SF}} = S_{\\mathrm{dB}} + \\mathrm{Norm}(0,\\sigma_{\\mathrm{SF}})$, and the paper claims the resulting CDF is well fitted by weighted Normal components for blocked and non-blocked line-of-sight states, with mean $\\mathbb{E}[S_{\\mathrm{dB}}]$ and standard deviation $\\sqrt{\\sigma_{\\mathrm{SF}}^2+\\sigma_{\\mathrm{SdB}}^2}$. Substituting this CDF into the MCS selection probabilities $m_j = F_S(s_{j+1}) - F_S(s_j)$ and then into the resource binning formula turns the session resource-request pmf into error functions. In the reported scenarios the fitted mean matches the exact mean SNR to the shown precision, the fitted standard deviation matches to three decimal places, and mean resource requests from the approximation differ from the full model by at most about two percent (e.g., 3.21 vs 3.27 at $p_C=0.1$ and 5 Mbps).","pith_inferences":["Because the Gaussian shape comes from convolution smoothing rather than from the specific UMi constants, the approximation likely carries over to other standardized deployment scenarios with different path-loss exponents and shadow-fading variances; testing this transfer is a direct next step.","The fixed interference margin leaves room for an SINR extension: a stochastic interference term could be absorbed into the Normal mixture instead of the constant $M_I$.","A spatially explicit simulation that conditions user distance on blocked versus unblocked state would test whether the distance-averaged $p_B$ in Eq. (16) is accurate or whether the correlation between blockage and distance shifts the mixture weights."],"forward_implications":["Queuing and session-level models of 5G NR can evaluate resource-request pmfs in closed form instead of numerical convolution.","Ignoring shadow fading is not a minor simplification: in the paper's tables the mean resource request without shadow fading is roughly 15 to 35 percent lower than with it.","The Normal fit gets better as shadow-fading variance increases, so the approximation is most reliable in the channels where direct numerical evaluation is most costly.","The same error-function machinery extends to random session rates, as the paper notes in its conclusion."],"supporting_citations":[{"why":"Supplies the 3GPP UMi street-canyon path-loss model, shadow-fading standard deviations, and blockage enhancements defining the channel.","marker":"[22]"},{"why":"Supplies the 3GPP NR physical-layer modulation and coding schemes and the SNR outage threshold used to size coverage.","marker":"[19]"},{"why":"Provides the SNR-to-MCS mapping table used to turn the SNR CDF into resource-request probabilities.","marker":"[23]"},{"why":"Gives the distance distribution for a uniformly located UE, the starting point of the SNR CDF derivation.","marker":"[24]"},{"why":"Provides the random-variable transformation technique used to convert distance density into SNR density.","marker":"[25]"},{"why":"Supplies the distance-dependent blockage probability formula that enters the mixture weight.","marker":"[26]"}],"fun_headline_variants":["5G mmWave SNR goes Gaussian, enabling closed-form resource estimates","Normal fit unlocks simple math for 5G mmWave resource requests","Error functions replace numerical integration in 5G mmWave resource modeling","Almost-exact Gaussian approximation for 5G mmWave resource requests","mmWave 5G: SNR CDF is practically Normal, simplifying resource pmf"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the blocked and non-blocked SNR states can be mixed with a single blockage probability averaged over all user distances, so the link state and the user's distance from the base station are treated as independent.","fun_headline_variants_meta":{"raw":{"variants":["5G mmWave SNR goes Gaussian, enabling closed-form resource estimates","Normal fit unlocks simple math for 5G mmWave resource requests","Error functions replace numerical integration in 5G mmWave resource modeling","Almost-exact Gaussian approximation for 5G mmWave resource requests","mmWave 5G: SNR CDF is practically Normal, simplifying resource pmf"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001089,"raw_usage":{"total_tokens":4558,"prompt_tokens":961,"completion_tokens":3597,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":577,"completion_tokens_details":{"reasoning_tokens":3502}},"tokens_in":577,"tokens_out":3597,"duration_ms":26094,"temperature":1.0,"reasoning_tokens":3502,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T11:28:18.348254+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Simulate a millimeter-wave sector with explicit blockers whose positions correlate with user distance, compute the empirical SNR CDF conditioned on blocked and unblocked states, and compare it with the paper's Eq. (16) mixture; if the Kolmogorov-Smirnov distance rises well above the reported 0.01 to 0.03, the independence assumption is the reason.","supporting_citations":[{"cited_title":"Study on channel model for frequencies from 0.5 to 100 GHz (Release 14),","cited_arxiv_id":null,"evidence_quote":"Supplies the 3GPP UMi street-canyon path-loss model, shadow-fading standard deviations, and blockage enhancements defining the channel."},{"cited_title":"NR; Physical channels and modulation (Release 15),","cited_arxiv_id":null,"evidence_quote":"Supplies the 3GPP NR physical-layer modulation and coding schemes and the SNR outage threshold used to size coverage."},{"cited_title":"MCS selection for throughput improvement in downlink LTE systems,","cited_arxiv_id":null,"evidence_quote":"Provides the SNR-to-MCS mapping table used to turn the SNR CDF into resource-request probabilities."},{"cited_title":"Distance distributions in random networks,","cited_arxiv_id":null,"evidence_quote":"Gives the distance distribution for a uniformly located UE, the starting point of the SNR CDF derivation."},{"cited_title":"Ross, Introduction to probability models","cited_arxiv_id":null,"evidence_quote":"Provides the random-variable transformation technique used to convert distance density into SNR density."},{"cited_title":"Capacity of Multiconnectivity mmWave Systems With Dynamic Blockage and Directional Antennas,","cited_arxiv_id":null,"evidence_quote":"Supplies the distance-dependent blockage probability formula that enters the mixture weight."}],"review_version":1}