{"id":"eb3949b1-87c4-47fc-805f-de323c9076ef","arxiv_id":"2411.15629","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":4.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"MILP-based planning selects optimal mixes of RIS, STAR RIS, NCR, and 3SNCR to minimize cost or maximize coverage in an urban mmWave network.","lead":"This paper proposes two optimization models for choosing where and what to install among smart radio devices to extend millimeter-wave coverage in a city. The models trade coverage against cost, and simulations indicate that adding newer devices can reduce deployment cost by up to 40%.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 40% cost-saving headline and optimal device mixes depend on the uniform 6 m building height in Sec. VI; real Milan height data could change SNR maps and deployment optima materially.","rationale":"The paper's strongest claim is really a conjunction: (i) the two MILP formulations produce optimal deployment plans, and (ii) adding STAR RIS and 3SNCR gives up to 40% cost savings. Part (i) is sound as mathematics: FCMC is a weighted set cover and MBCC a maximum coverage problem, and given the precomputed link-activation parameters, the constraints correctly enforce the coverage conditions. Part (ii) is an empirical result that depends on every input to the SNR computation. Among those inputs, the uniform 6 m building height is the most load-bearing because it is fixed in every experiment, it directly controls the deterministic LOS/blockage model, and it determines whether rooftop or wall-mounted devices are geometrically plausible. The cost model is also hand-assumed, but the authors vary the NCR/RIS price ratio in Fig. 6; no analogous sensitivity study is done for building heights. Real urban height variation can change which TPs are shadowed, which links meet the SNR threshold, and hence which devices and positions are selected, so the published savings and device-mix trends could be artifacts of a flat-city scenario. A rerun with real heights is a straightforward, decisive check. This is the same weak point the reader identified, so the conditional verdict remains appropriate; I would not strengthen or weaken it without the height sensitivity data.","tokens_in":16326,"tokens_out":10211,"duration_ms":101090,"concrete_test":"Re-run the pipeline for the eight Milan areas with actual building heights (e.g., from OpenStreetMap building:levels or a Milan 3D city model), keeping all Table I/II parameters unchanged. Regenerate the blockage/SNR maps and link-activation parameters, then solve FCMC and MBCC. Compare the average total costs and the full-vs-reduced-set savings with the paper's figures; in particular, check whether the 40% saving at M = 50 x 50 and the device contribution rankings survive. A material shift (e.g., more than 10 percentage points in savings or a different device mix) would show the quantitative claims depend on the height simplification; stability would resolve the concern.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section VI fixes every building at 6 m, with rooftop NCR/STAR RIS at 6.5 m and wall-mounted RIS at 5 m. Because the deterministic blockage and line-of-sight computation in Sec. III-B is built on this geometry, the link-activation parameters in (15)-(16) are all conditioned on a flat city. Actual Milan building heights vary from roughly 5 m to well over 30 m, which changes which TPs have direct BS coverage, which candidate sites are usable, and the relayed SNR for every device type. The FCMC/MBCC optimum is therefore only optimal for the assumed 6 m world. The conclusion's 'up to 40% cost savings' and the device-mix trends in Figs. 3-9 are quantitative claims drawn from that single height profile, and no sensitivity analysis is provided. The MILP framework itself is not invalidated; the concern is that the central numerical results are not established for realistic urban geometry.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops a planning framework for heterogeneous smart radio environments, in which a pre-deployed base station is augmented by RIS, STAR RIS, NCR, and 3SNCR devices. The authors formulate two MILPs: FCMC, which minimizes deployment cost subject to full coverage of all test points, and MBCC, which maximizes the number of covered test points under a budget constraint. The link activation parameters in (15)-(16) are computed from a Saleh-Valenzuela channel model with deterministic building blockage, using a map of Milan. Numerical results over eight 400 m x 400 m areas compare the reduced device set (RIS, NCR) with the full set (plus STAR RIS, 3SNCR), and report coverage-versus-budget curves, cost-versus-configuration trends, and the conclusion that advanced devices can yield up to 40% cost savings.","tokens_in":16547,"tokens_out":3481,"duration_ms":36247,"significance":"The MILP formulations are clear, correctly stated, and appropriate for the stated planning objectives; the FCMC and MBCC models are natural adaptations of set-cover and maximum-coverage problems, respectively. The paper's framework is genuinely useful because it treats device type, size, and placement jointly and is explicitly agnostic to specific pricing assumptions, so the optimization machinery can be reused as cost data improve. The numerical study is internally consistent, and the parameter tables in Table II make the experiments reproducible. The main value would be in providing planning insights for 6G deployment; however, the quantitative conclusions are conditioned on several strong simplifications, most notably a uniform building height, so the reported cost savings and optimal-device-mix trends should be read as illustrative rather than as established results for real urban environments.","major_comments":[{"comment":"The assumption that all buildings have a uniform height of 6 m is load-bearing for the quantitative results. This height enters the deterministic blockage calculation in Section III-B and therefore determines the link activation parameters in (15) and (16), which in turn fix the set of feasible deployment plans in the FCMC and MBCC models. It also determines which rooftop positions for NCR/STAR RIS and wall positions for RIS are usable. The conclusion that advanced devices achieve up to 40% cost savings, as well as the device-mix trends in Figures 3-9, are derived from this single height profile. Since no sensitivity analysis over building heights is provided, the central quantitative claims are not established for realistic urban geometry. I request either a sensitivity study with varied building heights or a explicit restatement of the numerical findings as illustrative for the assumed flat city.","section":"Section VI, first paragraph"},{"comment":"The paper claims that large realistic instances of FCMC and MBCC can be solved quickly without heuristics ('large and realistic planning instances can be solved quite quickly' and 'large instances can be solved with ease'), but no runtime data, instance sizes, solver information, or optimality gaps are reported. Since both formulations are NP-hard and the usefulness of the planning approach for real networks depends on scalability, this claim needs quantitative support. I ask for a table or plot reporting the numbers of candidate sites, test points, device configurations, solution times, and solver settings for the eight Milan scenarios.","section":"Sections V-B, V-C, and VI"},{"comment":"The cost model is hand-assumed: RIS cost scales linearly with the number of meta-atoms, NCR cost scales linearly with dB gain, STAR RIS is fixed at 2x RIS cost, and 3SNCR is priced identically to NCR. These assumptions directly affect the quantitative comparisons, including the 'up to 40% cost savings' headline and the optimal-RIS-size results in Figures 4 and 7. The Remark correctly states that the framework can accommodate updated prices, but it does not address how sensitive the conclusions are to the assumed relative prices. A robustness check that varies, for example, the STAR RIS multiplier and the NCR per-dB cost would strengthen the paper; alternatively, the quantitative claims should be explicitly framed as dependent on the listed pricing assumptions.","section":"Section IV, Table I and Remark"}],"minor_comments":[{"comment":"There are several typographical errors, e.g., 'central paradigms' should be 'central paradigm', 'muually' should be 'mutually', 'miigate' should be 'mitigate', 'architecures' should be 'architectures', and 'refer tosimultaneous' should be 'refer to simultaneous'.","section":"Abstract and Introduction"},{"comment":"The text says 'similarly to FMCM' but the correct acronym is FCMC; please fix this typo.","section":"Section V-C"},{"comment":"The notation xi in B^{2x1} is not defined; if B denotes the binary set, please state explicitly that xi is a binary selection vector.","section":"Section III-A, Eq. (9)"},{"comment":"The relationship between the NCR amplification gain |g|^2 = 55 dB, the NCR end-to-end gain G = 95 dB, and the text in Section II mentioning 95 dB should be clarified, as the two notations appear to refer to different quantities.","section":"Section II and Table II"},{"comment":"The caption reads 'vs RIS side size with, where the configurations and price...'; the comma and word order make the sentence ungrammatical and should be corrected.","section":"Figure 7 caption"}],"recommendation":"major_revision","confidential_remarks":"The core optimization framework is sound and the paper is worth publishing after revision. My main concern is that the quantitative conclusions, especially the 40% cost-savings claim, rest on a single assumed building-height profile and on hand-set cost multipliers, with no sensitivity analysis. The authors should either provide such analysis or clearly downgrade the numerical claims to illustrative. The missing runtime data for the scalability claim is also an important gap."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThis is a solid, incremental extension of the authors' own FCMC planning model. The new content is the inclusion of STAR RIS and trisectoral NCR alongside ordinary RIS/NCR, plus a budget-constrained coverage variant (MBCC). The MILP formulations are correct; FCMC is a set cover, MBCC a maximum coverage problem, and they say so. The channel models for each device are careful, and the cost scaling with RIS size and NCR gain is sensible.\n\nThe numerical study on the Milan map produces coherent trends: NCRs become more attractive as budget grows, and the advanced device set can reduce total cost by up to 40% in some scenarios. Those trends are internally consistent.\n\nThe main weakness is the geometry. Section VI assumes every building is 6 m tall. That single number determines which test points have line of sight, which candidate sites are usable, and what SNR each relay offers. The 40% savings and the optimal device mixes are computed for that flat 6 m world. Real Milan buildings vary from about 5 m to over 30 m, so the quantitative conclusions are not yet established for realistic urban geometry. This does not invalidate the framework—the optimization is agnostic to the inputs—but it does undercut the specific numerical claims. No sensitivity analysis is provided.\n\nSecond, the cost model is hand-assumed. The authors themselves call it a rational estimate, and they are right that relative prices drive the results. Still, the device-mix tipping points follow directly from those assumed relative prices. Fine as an illustrative example, but not a finding about real economics.\n\nThird, the paper claims large instances solve quickly, but no runtime data appear. For an MILP paper, that is a fixable gap.\n\nThe citation pattern is appropriate; the self-citations are to the prior FCMC paper and the authors' own channel work, which is fine here.\n\nThe paper is for operators and vendors planning mmWave deployments, and for researchers in RIS/NCR placement. The framework is usable and the trends are plausible, but the numbers should be read as conditional on the assumptions. I would accept this for peer review, with the expectation that the authors add a sensitivity study on building heights, report runtimes, and ideally release the simulation artifacts. If those come back solid, it becomes a useful reference.","headline":"Useful MILP extension for planning with STAR RIS and 3SNCR, but the headline savings are tied to a uniform 6 m city and hand-assumed costs.","tokens_in":17081,"tokens_out":2758,"would_cite":true,"duration_ms":24887,"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":"This paper claims that adding STAR RIS and tri-sectoral NCR devices to the network planner's toolkit can cut full-coverage deployment cost by up to 40% in dense urban mmWave scenarios, and proposes two MILP models to find the optimal…","keywords":["smart radio environment","network planning","reconfigurable intelligent surface","network-controlled repeater","STAR RIS","3SNCR","mixed-integer linear programming","millimeter-wave coverage"],"falsifier":"Take one of the eight Milan test areas, replace the uniform 6 m building height with actual heights from a city model, re-run the FCMC optimization with the full device set, and compare the chosen device mix and total cost; if the 40% savings figure or the dominance of STAR RIS and 3SNCR disappears, the planning insight is contingent on the height simplification.","tokens_in":16128,"feed_emoji":"📡","tokens_out":8703,"duration_ms":63072,"temperature":0.7,"pith_summary":"The paper asks how to deploy a heterogeneous set of smart radio environment devices—reflective intelligent surfaces (RIS), network-controlled repeaters (NCR), and their advanced variants STAR RIS and 3SNCR—to extend millimeter-wave coverage in a dense city either at minimum cost or within a fixed budget. It proposes two mixed-integer linear programs: FCMC, which minimizes total device cost while every test point must meet an SNR threshold, and MBCC, which maximizes the number of covered test points subject to a budget. Using a realistic map of Milan, a cluster-based channel model, building blockage, and costs that scale with RIS size and NCR gain, the paper shows that including the two advanced devices can lower deployment cost by up to 40% in some scenarios compared with conventional RIS and NCR alone. The underlying proposition is that network planning should choose among heterogeneous devices based on their physical coverage behavior, installation constraints, and relative price, not just on device count.","feed_headline":"Smart 6G devices cut urban coverage cost by up to 40%","feed_subtitle":"Two planning models pick the cheapest mix of reflectors and repeaters across eight Milan districts.","key_machinery":"The load-bearing object is the boolean link-activation parameter $\\Delta^d_{t,c}$, equal to 1 when a device of technology $d$ installed at candidate site $c$ yields SNR $\\geq \\Gamma$ at test point $t$, and 0 otherwise. These parameters are precomputed from the channel and blockage models, turning coverage into a set-cover constraint and letting the FCMC and MBCC formulations optimize over device types, configurations, and sites with a linear objective. The MILP optimality is what makes the numerical results planning insights rather than heuristic outcomes.","core_discovery":"The central claim is that a heterogeneous smart radio environment can be planned optimally by encoding each device's physical behavior and cost into boolean link-activation parameters, then solving a set-cover-style mixed-integer linear program. The paper proposes FCMC, which minimizes cost while covering all test points with signal-to-noise ratio above a threshold $\\Gamma$, and MBCC, which maximizes the number of covered test points while total device cost stays within a budget $B$. These models are evaluated on eight 400 m × 400 m urban areas in Milan at 28 GHz, with four device types, a Saleh-Valenzuela cluster channel model, deterministic building blockage, and a stochastic mobile-blocker model. The results indicate that STAR RIS and 3SNCR act as cost-effective substitutes: a rooftop STAR RIS can refract around buildings, and a 3SNCR covers a wider field of view than a two-panel NCR at comparable cost. In some scenarios the full device set achieves up to 40% lower total cost than the reduced set, and the optimal device mix shifts with budget, SNR threshold, RIS dimensions, NCR gain, and price ratio.","pith_inferences":["The MILP core is channel-agnostic: the same set-cover formulation should apply to other device families or frequency bands by recomputing the link-activation parameters and cost entries, so the planning framework is likely transferable beyond the four devices tested.","The 40% savings is an upper bound in the tested scenarios, not a universal constant; its size probably depends on how many test points sit in building shadows, a quantity the uniform-building-height assumption may overstate.","Because candidate sites are discrete, a practical next step would be to couple the MILP with a local continuous search over positions and orientations of the selected devices.","A direct validation would be to re-run the same optimization on one of the eight Milan areas using actual building heights from a digital city model; if the optimal device mix changes materially, height data should be treated as a first-class planning input."],"forward_implications":["In some urban scenarios, adding STAR RIS and 3SNCR to the device set reduces the total cost of full coverage by up to 40% compared with using only RIS and NCR.","Under a fixed budget, the optimal mix shifts toward NCRs and 3SNCRs as the budget grows, while STAR RISs take over serving users behind buildings through refraction.","An optimal RIS dimension exists, roughly 100×100 to 150×150 meta-atoms depending on the SNR threshold, beyond which larger surfaces cost more than the coverage they add.","Increasing NCR amplification gain beyond roughly 45–50 dB end-to-end raises cost without improving coverage; the optimal gain is lower for a 0 dB threshold than for 10 dB.","Requiring at least two devices per test point (K=2) raises cost substantially, but the full device set mitigates the increase, yielding a plan more robust to dynamic blockage."],"supporting_citations":[{"why":"Supplies the propagation and geometric comparison of NCR versus RIS in urban environments that underlies the SNR link-activation parameters and device-selection logic.","marker":"[24]"},{"why":"Provides the dynamic mobile-blocker model used to compute long-term SNR for direct and relayed links.","marker":"[29]"},{"why":"Introduces the FCMC model that this paper extends with additional device types and the MBCC formulation.","marker":"[30]"},{"why":"Provides the cost-scaling approach for RIS and NCR pricing used in the planning optimization.","marker":"[28]"},{"why":"Establishes the NP-hardness of the set-cover and maximum-coverage problems that the two MILP formulations solve.","marker":"[42]"},{"why":"Introduces the tri-sectoral NCR (3SNCR) configuration included in the full device set.","marker":"[10]"}],"fun_headline_variants":["Up to 40% cheaper 6G coverage via optimal device mix","Two planning algorithms pick best 6G reflector-repeater mix","Smart 6G planning cuts urban coverage cost by up to 40%","Optimize RIS and NCR placement to cut 6G costs in cities","Milan 6G test: new planning model saves up to 40%"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The paper assumes every building in the Milan test areas has the same height, 6 m, because precise height data were not available; this determines which test points have line of sight, where rooftop devices can sit, and how mmWave blockage is computed, so real height variation could change the optimal device types and positions.","fun_headline_variants_meta":{"raw":{"variants":["Up to 40% cheaper 6G coverage via optimal device mix","Two planning algorithms pick best 6G reflector-repeater mix","Smart 6G planning cuts urban coverage cost by up to 40%","Optimize RIS and NCR placement to cut 6G costs in cities","Milan 6G test: new planning model saves up to 40%"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000832,"raw_usage":{"total_tokens":3621,"prompt_tokens":924,"completion_tokens":2697,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":540,"completion_tokens_details":{"reasoning_tokens":2598}},"tokens_in":540,"tokens_out":2697,"duration_ms":17530,"temperature":1.0,"reasoning_tokens":2598,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T14:04:54.300681+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take one of the eight Milan test areas, replace the uniform 6 m building height with actual heights from a city model, re-run the FCMC optimization with the full device set, and compare the chosen device mix and total cost; if the 40% savings figure or the dominance of STAR RIS and 3SNCR disappears, the planning insight is contingent on the height simplification.","supporting_citations":[{"cited_title":"Network-controlled repeaters vs. reconfigurable intelligent surfaces for 6g mmw coverage extension: A simulative comparison,","cited_arxiv_id":null,"evidence_quote":"Supplies the propagation and geometric comparison of NCR versus RIS in urban environments that underlies the SNR link-activation parameters and device-selection logic."},{"cited_title":"The impact of mobile blockers on millimeter wave cellular systems,","cited_arxiv_id":null,"evidence_quote":"Provides the dynamic mobile-blocker model used to compute long-term SNR for direct and relayed links."},{"cited_title":"Advanced Network Planning in 6G Smart Radio Environments","cited_arxiv_id":"2411.06021","evidence_quote":"Introduces the FCMC model that this paper extends with additional device types and the MBCC formulation."},{"cited_title":"Planning Mm-Wave Access Networks With Reconfigurable Intelligent Surfaces","cited_arxiv_id":"2105.11755","evidence_quote":"Provides the cost-scaling approach for RIS and NCR pricing used in the planning optimization."},{"cited_title":"Korte and J","cited_arxiv_id":null,"evidence_quote":"Establishes the NP-hardness of the set-cover and maximum-coverage problems that the two MILP formulations solve."},{"cited_title":"Advanced tri-sectoral multi-user millimeter-wave smart repeater,","cited_arxiv_id":null,"evidence_quote":"Introduces the tri-sectoral NCR (3SNCR) configuration included in the full device set."}],"review_version":1}