Pith. sign in

REVIEW 4 major objections 6 minor 13 references

Indoor Sharing in the Mid-Band: A Performance Study of Neutral-Host, Cellular Macro, and Wi-Fi

T0 review · 4 major / 6 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read In a big-box store, a six-node CBRS neutral-host 4G network delivers higher median indoor throughput than outdoor macro 5G despite the macro using 225 MHz of bandwidth, and it covers the store with far fewer radios than enterprise Wi-Fi.

desk verdict Useful three-way measurement study, but the headline gains hinge on one MNO's neutral-host data and lack confidence intervals—indicative, not established. read the letter →

arxiv 2506.04974 v1 pith:FFES5AE4 submitted 2025-06-05 cs.NI eess.SP

Joshua Roy Palathinkal , Muhammad Iqbal Rochman , Vanlin Sathya , Mehmet Yavuz , Monisha Ghosh This is my paper
classification cs.NIeess.SP
keywords CBRSneutral-hostnetworkindoorcoveragemid-bandspectrumsharing5GmacroWi-Fi6buildingpenetrationlossfieldmeasurement
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper reports a side-by-side field measurement, inside a large big-box retail store, of three ways to deliver indoor mid-band connectivity: a CBRS neutral-host 4G network, the outdoor macro 4G/5G networks of the mobile operators, and an enterprise Wi-Fi 6 network. The central claim is that the neutral-host deployment wins on both coverage and throughput per unit of spectrum and infrastructure: six indoor CBSDs provide full coverage where enterprise Wi-Fi needs 65 access points, and the neutral-host's median indoor aggregated PHY downlink throughput is $2.08\times$ the macro 5G's despite using 40 MHz against 225 MHz. Per-channel normalized throughput is also higher indoors, with downlink gains of $1.44\times$ over 4G macro and $1.62\times$ over 5G macro, and uplink gains of $4.33\times$ and $13\times$. The paper argues this happens because indoor placement avoids building-penetration loss, with a measured median indoor-to-outdoor isolation of 26.6 dB, letting low-power indoor radios outperform distant outdoor towers. If correct, the result supports indoor-only sharing of mid-band spectrum as a credible alternative to dense Wi-Fi and outdoor macro for indoor traffic.

What carries the argument

The mechanism is the neutral-host CBRS b48 deployment: six ceiling-mounted CBSDs, each with two physical cell identities, transmitting at 24 dBm with 3 dBi omnidirectional antennas on five 20 MHz channels and supporting up to two-channel aggregation for a total of 40 MHz. The comparative tool that carries the argument is the normalized PHY throughput metric, expressed in bit/s/Hz/stream, which divides PDSCH/PUSCH throughput by the number of allocated resource blocks, subcarrier spacing, and MIMO layers, thereby isolating spectral efficiency from raw bandwidth so that a 40 MHz indoor 4G link can be compared fairly against a 225 MHz outdoor 5G link.

What would settle it

Run the same walk-test with an MNO-A-subscribed device on the same neutral-host network; if its median aggregated PHY downlink throughput is close to MNO-B's figure, the gains are not driven by throttling, but if it remains roughly 21 Mbps lower, the reported $2.08\times$, $4.33\times$, and $13\times$ gains are operator policy artifacts rather than neutral-host performance.

Watch

Extended reading notes

Core claim

The paper's central discovery is that an indoor CBRS neutral-host deployment, built from just six low-power ceiling-mounted CBSDs using at most 40 MHz of aggregated 4G bandwidth, provides both better indoor coverage and better indoor throughput than the outdoor macro 4G/5G networks it coexists with, and covers the building with roughly one tenth of the access points that the enterprise Wi-Fi deployment uses. In the same store, the neutral-host achieves a median indoor aggregated PHY-layer downlink throughput of $2.08\times$ the macro 5G network, even though that macro network aggregates up to 225 MHz across four 5G channels. The paper also reports per-channel normalized throughput gains of $1.44\times$ and $1.62\times$ in downlink and $4.33\times$ and $13\times$ in uplink against 4G and 5G macro respectively, and a $5.05\times$ application-layer HTTP downlink improvement over the indoor Wi-Fi 6 network. The measured median building loss of 26.6 dB is presented as both the reason for the neutral-host's indoor advantage and the mechanism that keeps its emissions from interfering with outdoor federal incumbents.

Load-bearing premise

The load-bearing premise is that MNO-B's neutral-host path, rather than MNO-A's throttled path, is the representative neutral-host performance; the paper excludes MNO-A because its subscribers saw 21 Mbps lower median PDSCH throughput, so the headline gains could reflect carrier policy instead of neutral-host capability.

Editorial extensions

If this is right

  • If the paper is correct, indoor coverage can be delivered with dramatically fewer radio nodes than Wi-Fi access points: six CBSDs matched the coverage that required 65 Wi-Fi APs.
  • Indoor neutral-host deployments can provide better uplink throughput than outdoor macro, with gains of $4.33\times$ and $13\times$, which matters for upload-heavy indoor applications.
  • Users on the neutral-host network transmit at lower uplink power than on macro networks, implying better device battery life and less uplink interference.
  • The 26.6 dB median building isolation means indoor CBRS radios can share spectrum with outdoor incumbents while maintaining strong indoor coverage, supporting the feasibility of indoor-only shared spectrum.
  • Deployment location inside the building can matter more than total spectrum bandwidth, since a 40 MHz indoor network beat a 225 MHz outdoor network on median indoor downlink throughput.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Editorial inference: the same indoor-placement logic should transfer to the newly proposed shared bands, such as 3.1–3.45 GHz and 7.125–8.4 GHz, where building loss could protect outdoor incumbents while giving indoor users a coverage advantage over outdoor macro.
  • A testable extension would be to compare the neutral-host against an outdoor small cell using the same CBRS bands, which would separate the benefit of indoor placement from the benefit of CBRS-specific spectrum properties.
  • If operator throttling on the MNO-A path were removed, the paper's own logic predicts that MNO-A's neutral-host throughput would rise to match MNO-B's; a measurement with an unthrottled MNO-A subscriber would confirm that the observed gap is policy, not radio physics.
  • The Wi-Fi comparison used 20 MHz channels and AP transmit powers of 10–19 dBm; a modern Wi-Fi 6E deployment with wider channels could narrow the measured $5.05\times$ gap, so the comparison is specific to the configured Wi-Fi network rather than to Wi-Fi in general.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 6 minor

Summary. This paper reports an in-situ measurement campaign in a single large big-box retail store, comparing three co-located mid-band wireless systems: a CBRS-based neutral-host (NH) 4G/5G network with six indoor CBSDs, the outdoor macro 4G/5G networks of three MNOs, and an enterprise Wi-Fi 6 network with 65 identified 5 GHz BSSIDs. The central claims are that the NH deployment achieves full indoor coverage with far fewer infrastructure nodes than Wi-Fi, that its per-channel normalized throughput beats indoor 4G/5G macro by 1.44x/1.62x in downlink and 4.33x/13x in uplink, that its aggregated indoor PHY-layer downlink throughput is 2.08x higher than a 5G macro deployment despite using only 40 MHz versus 225 MHz, and that its application-layer HTTP downlink throughput is 5.05x higher than Wi-Fi. The paper is purely empirical and draws policy implications for future indoor shared-spectrum bands.

Significance. If the quantitative claims are reliable, the results are significant for indoor spectrum sharing and neutral-host deployment: a low-power indoor CBRS deployment may outperform outdoor macro cellular with far less spectrum and infrastructure, and may require many fewer indoor radio nodes than Wi-Fi. The study has genuine strengths: a large real-world dataset (221,396 cellular and 570,803 Wi-Fi datapoints), simultaneous measurement of three co-located technologies, documentation of deployment parameters, quantification of building loss (26.6 dB median), and use of a normalized spectral-efficiency metric that partially controls for bandwidth and MIMO configuration. However, the headline ratios are median estimates from a single walking campaign, without confidence intervals, effective sample sizes, or statistical tests, and the NH analysis is restricted to one MNO after observing a throughput difference attributed to throttling. As a result, the qualitative direction of the findings is plausible and consistent with propagation physics, but the specific numerical gains are not established at the precision implied by the abstract.

major comments (4)
  1. [Section IV-B] The restriction of the neutral-host analysis to MNO-B is load-bearing and is not adequately justified. The text states that a MNO-B-subscribed device achieved 21 Mbps higher median PDSCH throughput than a MNO-A-subscribed device, attributes this to operator throttling, and then says, "we therefore focus exclusively on the performance analysis of MNO-B's NH deployment throughout this study." No evidence is provided that distinguishes MNO-A throttling from other differences between the two NH paths, such as CBSD cell load, PCI assignment, or coverage quality. Since every headline throughput gain (Figs. 4 and 7) is computed on MNO-B's NH data only, the reported 2.08x, 4.33x, and 13x ratios could reflect MNO-specific routing policy rather than neutral-host capability. Please include the MNO-A NH data in a sensitivity analysis, or provide direct evidence of throttling from the infrastructure provider or operator policy.
  2. [Section IV-B, Figs. 4 and 7] The median ratios that support the abstract's quantitative claims are presented without confidence intervals, sample sizes, or statistical tests. The 221,396 total datapoints span multiple bands, MNOs, QualiPoc report types, and indoor/outdoor directions, so the effective sample size underlying any single box in Figs. 4 and 7 is not reported and is likely much smaller than the total. A walking measurement campaign produces spatially and temporally correlated samples, and the medians are sensitive to route, dwell time, and scheduling fairness. Please report per-technology, per-direction, and per-location sample counts, compute bootstrap confidence intervals for each ratio, and test whether the indoor NH-versus-macro differences remain significant after accounting for spatial correlation.
  3. [Sections IV-A and IV-C] The coverage claim that six CBSDs achieve "full coverage" where enterprise Wi-Fi requires 65 APs is not operationalized. "Full coverage" is never defined in terms of a coverage threshold (e.g., a minimum RSRP or RSSI value) or a coverage area fraction, so the comparison cannot be verified or reproduced. In addition, the paper equates 65 unique 5 GHz BSSIDs with 65 APs; if a physical AP has multiple BSSIDs per radio or if the 2.4 GHz BSSID count of 12 is taken into account, this equivalence is not automatic. Please define the coverage criterion, apply it identically to the NH and Wi-Fi deployments, and report the resulting coverage outcomes for the actual six-CBSD and observed-BSSID deployments.
  4. [Sections II and IV-D] The application-layer throughput comparison is confounded by the use of different test endpoints for the two directions: HTTP downlink uses github.com while HTTP uplink uses httpbin.org. These hosts have different CDNs, server locations, and load characteristics, which can produce throughput differences unrelated to the access network. The text itself attributes the anomalous Wi-Fi result of higher uplink than downlink throughput to "the difference in target host utilized in the HTTP GET and PUT request." This confound affects the comparability of the HTTP-layer results, including the 5.05x Wi-Fi downlink claim and the 1.92x uplink claim. Please use a single controllable server for both directions, or provide per-endpoint calibration measurements.
minor comments (6)
  1. [Section II] The HTTP test uses 5-second transfers; at the throughput levels reported (hundreds of Mbps), 5 seconds may not reach steady state. Please report session durations and consider whether the short test length biases the application-layer medians.
  2. [Section IV-C] The statement that the 5 GHz band "has more coverage" is based on the percentage of time the UE was connected to each band, which is a channel-occupancy metric rather than a coverage metric. Please relabel this as connected-time share or provide an RSSI-threshold-based coverage map.
  3. [Section IV-B] The normalized throughput metric is referenced to prior work [11] but its formula is not reproduced. Because this metric is central to the 1.44x-13x claims, please include the exact definition (including the treatment of resource blocks, subcarrier spacing, and MIMO layers) in the paper or an appendix.
  4. [Table III and Section IV-D] The paper states that the NH deployment supports aggregation of up to two 20 MHz channels (40 MHz total) and that Fig. 7a compares aggregated throughput, but it does not report how often two-channel aggregation actually occurred during the measurements. Since the 2.08x claim depends on aggregated throughput, this statistic should be reported.
  5. [Section IV-D] The text states that MNO-B's 5G macro network operates in "5G standalone mode," but no evidence for standalone versus non-standalone operation is given. Please clarify how this was determined and whether it affects the comparison.
  6. [General] No data or code release is mentioned. Given that the central claims are empirical medians from a single campaign, making the anonymized measurement data and analysis code available would substantially strengthen reproducibility and allow independent verification of the reported ratios.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found: all headline claims are direct empirical comparisons of measured data, with no fitted parameters, no self-citation chain, and no derivation that reduces to its own inputs.

full rationale

This paper reports a direct measurement campaign; every headline quantity is a median of captured data (RSRP, PDSCH/PUSCH throughput, normalized throughput, HTTP throughput) from three co-located deployments, and no model parameter is fitted to any subset of the data in order to 'predict' another subset. The only externally sourced object is the normalized-throughput metric of [11], which is cited as a definition (bit/s/Hz/stream) and applied identically to NH and macro samples; a metric definition that does not contain the target result is not load-bearing circularity. The MNO-B-only restriction ('we focus exclusively on the performance analysis of MNO-B's NH deployment') is a data-selection choice motivated by observed MNO-A throttling; it may affect representativeness, but it does not define the claimed gains into existence, since the 2.08x, 4.33x, 13x, and 5.05x figures still require the independently measured macro and Wi-Fi samples to be lower. Self-citations [7]-[9] and [11] are contextual prior work or metric references; no central claim is justified solely by a self-citation whose content is unverified. Accordingly no circular step reduces a prediction to its inputs by construction.

Assumptions & free parameters 0 free parameters · 4 assumptions · 0 invented entities

No fitted parameters or derived model; the paper's contribution is empirical. Its claims rest on the measurement assumptions listed above.

assumptions (4)
  • domain assumption QualiPoc RSRP/PDSCH/PUSCH and SigCap RSSI/beacon readings accurately characterize the radio environment at each walk-through location.
    All central measurements rely on consumer phone and vendor tools; no calibration or validation against reference equipment is reported (Sections II, IV).
  • domain assumption MNO-B's NH data is representative of the NH deployment's capability.
    MNO-A subscriber data were set aside after observing throttling-like lower throughput; the paper states this is done for fairness, but it assumes the unthrottled operator path is the right yardstick (Section IV-B).
  • domain assumption HTTP downloads/uploads to github.com and httpbin.org provide a stable application-layer throughput reference.
    Application-layer comparison depends on external servers and internet paths that are not controlled; the paper itself notes server differences between DL and UL (Section IV-D).
  • domain assumption Comparing RSRP heatmaps (NH/macro) with RSSI heatmaps (Wi-Fi) is a valid coverage equivalence test.
    The coverage claim of six CBSDs versus 65 APs rests on unstated comparability of different signal metrics and no threshold definition of full coverage (Sections IV-A and IV-C).

how reviews work

0 comments
Cite this review

Pith. "Pith review of Indoor Sharing in the Mid-Band: A Performance Study of Neutral-Host, Cellular Macro, and Wi-Fi." pith.science (2026). https://pith.science/paper/FFES5AE4

@misc{pith2026250604974,
  author       = {Pith},
  title        = {Pith review of: Indoor Sharing in the Mid-Band: A Performance Study of Neutral-Host, Cellular Macro, and Wi-Fi},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/FFES5AE4}},
  note         = {Machine review of arXiv:2506.04974}
}
read the original abstract

Indoor environments present a significant challenge for wireless connectivity, as immense data demand strains traditional solutions. Public Mobile Network Operators (MNOs), utilizing outdoor macro base stations (BSs), suffer from poor signal penetration. Indoor Wi-Fi networks, on the other hand, may face reliability issues due to spectrum contention. Shared spectrum models, particularly the Citizens Broadband Radio Service (CBRS) utilized by private 4G/5G networks, have emerged as a promising alternative to provide reliable indoor service. Moreover, these private networks are equipped with the neutral-host (NH) model, seamlessly offloading indoor MNOs' traffic to the private CBRS network. This paper presents a comprehensive, in-situ performance evaluation of three co-located technologies utilizing mid-bands spectrum (1-6 GHz)--a CBRS-based NH network, public MNO macro networks, and a Wi-Fi 6 network--within a large, big-box retail store characterized by significant building loss. Our analysis demonstrates: (i) the NH network provides superior indoor coverage compared to MNO macro, requiring only six CBRS devices (CBSDs)--versus 65 Access Points (APs) for enterprise Wi-Fi--to achieve full coverage, with a median building loss of 26.6 dB ensuring interference-free coexistence with outdoor federal incumbents; (ii) the NH network achieves substantial indoor throughput gains, with per-channel normalized throughput improvements of 1.44x and 1.62x in downlink (DL), and 4.33x and 13x in uplink (UL), compared to 4G and 5G macro deployments, respectively; (iii) the NH deployment achieves a median indoor aggregated physical (PHY)-layer DL throughput gain of 2.08x over 5G macro deployments indoors, despite utilizing only 40 MHz of aggregated bandwidth compared to 225 MHz for 5G macro; and (iv) the NH deployment also outperforms Wi-Fi in application-layer HTTP DL performance by 5.05x.

Figures

Figures reproduced from arXiv: 2506.04974 by the authors.

Figure 1
Figure 1. RSRP heatmap comparison across NH and MNO macro deployments. [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Coverage statistics for representative CBSD/PCI. [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. PHY-layer throughput comparison across NH and MNO macro deployments. [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: PHY-layer performance analysis of MNO-B’s NH and macro deployments. [PITH_FULL_IMAGE:figures/full_fig_p004_4.png]
Figure 5
Figure 5. Figure 5: Wi-Fi coverage maps and time-based band usage statistics. [PITH_FULL_IMAGE:figures/full_fig_p005_5.png]
Figure 6
Figure 6. Figure 6: Number of unique BSSIDs across 5 GHz channels [PITH_FULL_IMAGE:figures/full_fig_p005_6.png]
Figure 7
Figure 7. Figure 7: User-experienced throughput statistics. Fig. 7b shows the indoor DL performance of MNO-B’s NH outperforms enterprise Wi-Fi with a 5.05× improvement. Further comparisons with MNO-B’s macro deployments cor￾roborate the aggregated PHY-layer PDSCH throughput results presen…

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

13 extracted references · 13 canonical work pages

  1. [1]

    Ericsson Mobility Report,

    Ericsson, “Ericsson Mobility Report,” Retrieved from https: //www.ericsson.com/4ae12c/assets/local/reports-papers/mobility-report/ documents/2023/ericsson-mobility-report-november-2023.pdf, Nov. 2023, accessed: May 2025

    work page 2023

  2. [2]

    The next frontier in spectrum policy: Indoor-only sharing of federal bands,

    M. Calabrese and J. Dine, “The next frontier in spectrum policy: Indoor-only sharing of federal bands,” Retrieved from https://d1y8sb8igg2f8e.cloudfront.net/documents/The Next Frontier in Spectrum Policy 2024-11-25 141303 RP4Nq3g.pdf, Nov. 2024, accessed: May 2025

    work page 2024

  3. [3]

    Evaluating the interference potential in 6 GHz: An extensive measurement campaign of a dense indoor Wi-Fi 6E network,

    S. Dogan-Tusha, M. I. Rochman, A. Tusha, H. Nasiri, J. Helzerman, and M. Ghosh, “Evaluating the interference potential in 6 GHz: An extensive measurement campaign of a dense indoor Wi-Fi 6E network,” in Proceedings of the 17th ACM Workshop on Wireless Network Testbeds, Experimental evaluation & Characterization , 2023, pp. 56–63

    work page 2023

  4. [4]

    Indoor and outdoor measurement campaign for unlicensed 6 GHz operation with Wi-Fi 6E,

    S. Dogan-Tusha, A. Tusha, H. Nasiri, M. I. Rochman, and M. Ghosh, “Indoor and outdoor measurement campaign for unlicensed 6 GHz operation with Wi-Fi 6E,” in 2023 26th International Symposium on Wireless Personal Multimedia Communications (WPMC) . IEEE, 2023, pp. 1–6

    work page 2023

  5. [5]

    A comprehensive analysis of secondary coexistence in a real-world CBRS deployment,

    A. Tusha, S. Dogan-Tusha, J. R. Palathinkal, H. Nasiri, M. I. Rochman, P. McGuire, and M. Ghosh, “A comprehensive analysis of secondary coexistence in a real-world CBRS deployment,” IEEE Transactions on Cognitive Communications and Networking , 2025

    work page 2025

  6. [6]

    Neutral host technology: The future of mobile network operators,

    R. Bajracharya, R. Shrestha, H. Jung, and H. Shin, “Neutral host technology: The future of mobile network operators,” IEEE Access , vol. 10, pp. 99 221–99 234, 2022

    work page 2022

  7. [7]

    Neutral host deployment-a mea- surement study on routing and mobility management between public and private network,

    V . Sathya, M. Shah, and M. Yavuz, “Neutral host deployment-a mea- surement study on routing and mobility management between public and private network,” in 2023 IEEE Future Networks World F orum (FNWF). IEEE, 2023, pp. 1–8

    work page 2023

  8. [8]

    Warehouse deployment: A comparative measurement study of commercial Wi-Fi and CBRS systems,

    V . Sathya, L. Zhang, M. Goyal, and M. Yavuz, “Warehouse deployment: A comparative measurement study of commercial Wi-Fi and CBRS systems,” in 2023 International Conference on Computing, Networking and Communications (ICNC) . IEEE, 2023, pp. 242–248

    work page 2023

Show all 13 references
  1. [9]

    Neutral-hosts in the shared mid-bands: Addressing indoor cellular performance,

    M. I. Rochman, J. R. Palathinkal, V . Sathya, M. Yavuz, and M. Ghosh, “Neutral-hosts in the shared mid-bands: Addressing indoor cellular performance,” 2025. [Online]. Available: https: //arxiv.org/abs/2505.18360

    2025

  2. [10]

    The National Spectrum Strategy,

    National Telecommunications and Information Administration, “The National Spectrum Strategy,” Retrieved from https://www.ntia.gov/sites/ default/files/publications/national spectrum strategy final.pdf, 2023, accessed: Dec. 2024

    2023

  3. [11]

    A comprehensive real-world evaluation of 5G improvements over 4G in low-and mid- bands,

    M. I. Rochman, W. Ye, Z.-L. Zhang, and M. Ghosh, “A comprehensive real-world evaluation of 5G improvements over 4G in low-and mid- bands,” IEEE Transactions on Cognitive Communications and Network- ing, 2025

    2025

  4. [12]

    QualiPoc Android,

    Rohde & Schwarz, “QualiPoc Android,” Retrieved from https://www.rohde-schwarz.com/us/products/test-and-measurement/ network-data-collection/qualipoc-android \ 63493-55430.html, accessed: Dec. 2024

    2024

  5. [13]

    Measurement-based coexis- tence studies of LAA & Wi-Fi deployments in chicago,

    V . Sathya, M. I. Rochman, and M. Ghosh, “Measurement-based coexis- tence studies of LAA & Wi-Fi deployments in chicago,” IEEE Wireless Communications, vol. 28, no. 1, pp. 136–143, 2020

    2020

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.