Pith. sign in

REVIEW 3 major objections 6 minor 46 references

Cooperative Mapping, Localization, and Beam Management via Multi-Modal SLAM in ISAC Systems

T0 review · 3 major / 6 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read A unified multi-modal SLAM loop ties cooperative radio map building, camera/IMU fusion, and beam management together, claiming up to 60% better maps, 37.5% sharper localization, and a 36% spectral-efficiency gain in mmWave ISAC systems.

desk verdict Solid integration of known radio-SLAM and beam-management pieces, but the evidence is simulation-only and the IUI-rejection claim needs a closer look at its proxy. read the letter →

arxiv 2507.05718 v1 pith:WBN4KT2O submitted 2025-07-08 cs.IT math.IT

Hang Que , Jie Yang , Tao Du , Shuqiang Xia , Chao-Kai Wen , Shi Jin This is my paper · ORCID
classification cs.ITmath.IT
keywords integratedsensingandcommunicationmulti-userSLAMradiomapconstructionbeammanagementmulti-modallocalizationmmWaveinter-userinterferencevirtualanchor
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

The paper is trying to show that one feedback loop can serve sensing and communication: multiple UEs jointly build a radio map, the map improves their own positions, and the map-plus-positions tell the base station which beams to use. A sympathetic reader should care because 6G millimeter-wave beam management currently wastes overhead on sweeping and is blind to which reflectors matter, while camera and IMU data are usually treated as separate from radio SLAM. The paper claims a unified Bayesian treatment of cooperative multi-user SLAM, with a two-stage map-building algorithm, is robust enough to handle dynamic visibility and features that appear or disappear, and that the same sensing outputs transfer directly into beam priors that cut inter-user interference. If true, this is a path to making sensing a communication asset rather than an add-on.

What carries the argument

The load-bearing object is the global radio map Q(t), a set of virtual anchors — mirror images of the base station in reflective walls — each with a location, covariance, visibility vector across UEs, confidence, and last-update timestamp. The argument works through a two-stage Bayesian factor-graph SLAM where UEs run belief-propagation particle filters locally, the base station fuses uploaded local feature sets into Q(t) with nearest-neighbor association and variance-weighted updates, and UEs then download Q(t) as prior features to refine their own maps. The camera branch contributes a pixel-position-dependent variance lookup table, and beam management converts map features visible to each UE into ordered path priors (angles, search range, gain), selects paths while enforcing a minimum angular separation, and builds tracking codebooks around those priors.

What would settle it

Run the outdoor CARLA scenario with a second vehicle crossing between the true UE and the camera so that the nearest detection in Eq. (21) is the wrong car; if the fused localization error rises above non-fused multi-user SLAM for more than occasional frames, the camera-fusion claim fails. A simpler check is to recompute the reported 37.5% localization gain while varying the association threshold DTh across 0.5 to 2 meters and see whether the gain survives.

Watch

Extended reading notes

Core claim

The paper proposes a single Bayesian framework in which multiple UEs run local radio SLAM from millimeter-wave angle estimates and IMU predictions, upload local virtual-anchor features to the base station, and then download a refined global radio map that in turn improves local estimation. The base station's stereo camera detections are fused with SLAM and IMU outputs through a pixel-dependent error model to produce UE locations, and those locations plus the map are converted into per-UE beam priors. The claimed payoff is a 60% reduction in radio-map OSPA error over single-user SLAM, a 37.5% improvement in UE localization accuracy over single-modal SLAM, and a 36% higher spectral efficiency than beam tracking with the same measurement budget.

Load-bearing premise

The whole vision gain rests on the assumption that the closest YOLO-detected object to each SLAM-tracked UE is that UE and that the offline-trained pixel-error table remains valid; if a nearer non-UE object or an invalid variance estimate attaches to the UE, the fused position is corrupted rather than improved.

Editorial extensions

If this is right

  • If the central claim holds, radio maps can be reconstructed collaboratively by ordinary UEs in deployment, reducing the need for site surveys or dedicated sensing infrastructure.
  • The map and localization outputs supply beam priors, so beam training can concentrate measurements on a narrow angular region rather than sweeping the full codebook, keeping overhead constant while improving resolution.
  • The inter-user interference mitigation step implies that spectral efficiency can improve as the number of UEs grows, in contrast to conventional beam sweeping where denser UEs suffer more overlapping beams.
  • Multi-modal fusion compensates for each sensor's individual failure mode: YOLO missed detections and IMU drift are bounded by the other modalities, so localization remains continuous in time.
  • The two-stage design can adapt to feature birth, feature death, and heterogeneous fields of view, suggesting the map does not need to be re-initialized when UEs lose or gain visibility of reflectors.

Reading between the lines

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

  • The paper leaves open whether the nearest-neighbor camera-to-UE association in Eq. (21) survives dense scenes; a natural extension is probabilistic data association or track-based association, which would make the vision branch more robust when a non-UE object appears closer than the true UE.
  • The offline-trained pixel-error lookup table assumes the camera pose and scene geometry stay fixed; a testable extension is to re-estimate the table periodically or condition it on scene depth, because an outdated table would bias the fused estimate exactly where the camera is most needed.
  • Because the global radio map carries timestamps and visibility vectors, the same machinery could plausibly support predictive beam handover as UEs move between cells or around blockages, a use case the paper does not explore.
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

3 major / 6 minor

Summary. This paper proposes a multi-modal SLAM framework for multi-user mmWave ISAC systems, combining cooperative radio SLAM, camera and IMU fusion, and sensing-aided beam management in one pipeline. The authors formulate a Bayesian factor-graph model that separates UE motion, local map transition, measurement update/data association, and global map update, and implement it as a two-stage algorithm (initialization without a global map, then refinement with global-map feedback). Localization fuses the radio/IMU SLAM estimate with YOLO/stereo-camera detections through a nearest-neighbor association gate and an offline-trained pixel-dependent error model. Beam management generates per-UE path priors from the global radio map and UE positions, switches selected paths based on a predicted spectral efficiency criterion intended to reduce inter-user interference, and then performs codebook-based tracking. Simulations in an indoor 10x10 m room and an outdoor CARLA street report radio-map OSPA improvements up to 60%, a windowed localization error reduction of 37.5%, and downlink SE gains of 36% and 149% over tracking and sweeping baselines.

Significance. The main strength of the paper is the breadth of the integration: a single Bayesian framework supports three concrete algorithmic outputs (map, location, beam selection), and the simulations compare against reasonable external baselines (single-user SLAM, AA and GCI fusion, tracking and sweeping beam management) rather than only against ablated versions of the proposed method. The outdoor CARLA evaluation with a real YOLOv8 detector and an offline-trained error lookup table is a concrete, falsifiable validation setup, and the paper is candid about the limitations of vision-only localization. If the results hold, the framework is a useful demonstration that SLAM-derived sensing can be converted into communication-layer gains in multi-user scenarios. The main reservations are that all quantitative claims rest on single synthetic scenarios without confidence intervals, the headline localization improvement is confined to a short time window, and the beam-management decision rule is not analyzed beyond the chosen scenario; these are addressed in the major comments.

major comments (3)
  1. [Abstract; Section VII-B and Fig. 9(b)] The abstract claims that the proposed system 'enhances localization accuracy by 37.5%' without qualification, but Section VII-B reports this figure only for the time interval t in (15, 40) s in the outdoor scenario, when vehicles are near the middle of the street where the stereo camera is most accurate; the text states that multi-modal localization outperforms the other two SLAM algorithms only when t in (15, 40) s, implying the advantage does not persist outside this window. Please report the full-horizon average localization error with confidence bounds for the multi-user and multi-modal curves, and qualify the headline claim in the abstract and conclusion with the window over which it holds.
  2. [Section VI, Algorithm 4, Eqs. (24)-(26)] The switching step in Algorithm 4 (lines 6-14) maximizes the predicted spectral efficiency of Eq. (26), evaluated with the estimated channel H_j(t) built in Eq. (24) from the prior paths selected for UE j. Because the interference term in Eq. (26) sums |w_j^H H_j(t) f_i(t)|^2 over i not equal to j, it is near zero whenever UE j's selected paths and UE i's chosen beam have disjoint angular supports; in the extreme case of disjoint visibility sets, the interference term is identically zero and the switching logic can only react to the UE's own path gains. The 36% gain in Fig. 6(b) is computed with the true SE of Eq. (6) as stated in Section VII, which mitigates the concern that the reported number itself is a predicted-SE artifact, but the paper still provides no analysis of how well the predicted SE tracks the true SE after a switch, no ablation isolating the contribution of the switching step from the tracking codebooks, and no discussion of the degenerate disjoint-visibility case. Please add such an analysis or ablation and state the conditions under which the predicted-SE proxy is reliable.
  3. [Section V, Eqs. (20)-(22); Section VII-B] The claimed localization gain rests on two assumptions that are not quantified. First, the nearest-neighbor association in Eq. (21) is never analyzed for misassociation probability, yet assigning the wrong camera detection to a UE corrupts the fused estimate in Eq. (22); the gating by the threshold D_Th does not by itself guarantee correct association when multiple objects are present. Second, the error lookup table in Eq. (20) is trained offline and validated only by a mean deviation of 0.036 m in CARLA, and is assumed to describe the vision error at all pixel locations and scenarios, including the online deployment conditions; Section VII-B itself shows that vision-only localization is discontinuous because of YOLO missed detections, so the camera branch is intermittently inactive. Please quantify the sensitivity of the localization improvement to misassociation rates, to the threshold D_Th, and to lookup-table mismatch, or the attribution of the 37.5% gain to the camera fusion branch remains unsupported.
minor comments (6)
  1. [Section VII] All headline percentages (60%, 37.5%, 36%, 149%) are reported without confidence intervals or error bars even though 1,000 Monte Carlo runs are performed; adding standard-error bands to Figs. 4-6 and 9 would substantiate the comparisons.
  2. [Section VII-A, footnote 4] The OSPA curves for scenarios 2 and 3 are omitted 'due to space limits'; the claim that the algorithm maintains comparable radio map accuracy under local feature birth/death should be supported by the actual curves or by a quantitative statement of the maximum degradation relative to Scenario 1.
  3. [Section VII-B] The outdoor beam management analysis is stated to be omitted 'due to similar results', but no figure or numerical summary is given; the abstract's claim of significant outperformance in 'both indoor and outdoor environments' would benefit from at least a brief outdoor SE comparison.
  4. [Section IV-A and Section VII-A] The parameter r_cut is used as a pruning/confidence threshold in Eqs. (14)-(17) but is described as a 'data association distance threshold' in Section VII-A, while Algorithm 1 uses d_th for the association distance; the parameter naming should be made consistent in the simulation description.
  5. [Section III and Section IV-A] The pruning criterion in Eq. (16) uses a decay factor 2^(t_cand - t) with no stated rationale or time-unit convention, and the sentence 'we assume the radio map states as Q(0) = empty, P1:J = empty' appears to mean P_{1:J}(0) = empty; a clarification of both would help.
  6. [Algorithms 1-2 and throughout] Algorithm 2 lines 7 and 17 refer to updates 'similar to (16)', but (16) is the pruning criterion; the intended reference appears to be the fusion equations (15a)-(15e). Several minor typos also occur, such as inconsistent spacing 'D VIS-RF&IMU' versus 'DVIS-RF&IMU' in and around Algorithm 3, and interchangeable 'VA2', 'V A2', and 'V A 2' in Section VII.

Circularity Check

0 steps flagged · score 2.0 of 10

No significant circularity: the paper's mapping, localization, and beam-management results are validated against independent baselines and do not reduce to their own inputs.

full rationale

The paper's central claims are evaluated externally rather than derived from the assumptions. Radio-map accuracy is compared against single-user SLAM, AA, and GCI in simulation with ground-truth virtual anchors, and localization is compared against IMU-only, vision-only, single-user SLAM, and multi-user SLAM using true UE trajectories. The vision error model in Eq. (20) is fitted offline to a separate training set and validated on a disjoint validation set, so the 37.5% localization improvement is a measured simulation outcome rather than a restatement of the fitted variances. The beam-management algorithm internally optimizes the 'predicted SE' in Eq. (26) using the estimated channel of Eq. (24), but the reported 36% SE gain is computed from the true channel via Eq. (6), so the internal optimization proxy is a correctness/robustness concern, not a circular reduction. The paper does cite the authors' own prior work [1], [12], and [23] for the factor-graph structure, IMU noise, and azimuth search-range heuristic, but these citations supply reusable components and measurement parameters, not the paper's claimed results, and the main contributions are benchmarked against external methods. No step in the derivation chain is equivalent to its inputs by construction.

Assumptions & free parameters 7 free parameters · 5 assumptions · 0 invented entities

No new physical entities are introduced; virtual anchors, visibility vectors, and global radio map states are modeling constructs from prior SLAM literature. The central claim depends on seven hand-chosen or fitted parameters, several of which are not specified numerically, and on the domain assumptions listed above.

free parameters (7)
  • Stage switch time T_config = 10 s
    Chosen so local SLAM converges before global map refinement; directly controls when Algorithm 1 gives way to Algorithm 2.
  • Confidence threshold r_cut = 0.3
    Used in feature pruning and visibility gating in Eqs. (14), (16), (17), and (18); set by hand and stated to scale with scenario size.
  • Association distance threshold d_th = not specified in paper
    Decides whether a local feature merges with an existing global feature in Algorithms 1 and 2; the exact value is omitted.
  • Vision-SLAM association threshold D_th = not specified in paper
    Gates camera fusion in Algorithm 3; without it the fusion rule in Eq. (22) cannot be reproduced.
  • Interference angle threshold theta_th = not specified in paper
    Triggers beam switching in Algorithm 4; the value affects the reported anti-interference gain.
  • Image block sizes u_cell, v_cell = not specified in paper
    Determine the resolution of the vision error look-up table in Eq. (20).
  • Offline vision error look-up table e_VISX/Y = trained on 1,000 CARLA vehicles
    Fitted per-block average errors used as variance weights in the fusion equation; fitted to simulator data, not real camera data.
assumptions (5)
  • domain assumption Single-bounce specular reflection model for NLoS paths
    Section II-A and Eq. (4): every NLoS path is a single reflection off a flat wall producing a virtual anchor; paths with two or more reflections are absorbed into noise. The radio map and beam priors depend on this representation.
  • domain assumption Gaussian measurement noise for angles, IMU positions, and camera positions
    Eqs. (7), (13), and (19): all sensors are modeled with known Gaussian variances, which justifies the product-form fusion in Eq. (22). Real angle and vision errors may be non-Gaussian and correlated.
  • standard math Markov state-transition and conditional independence in the Bayesian posterior
    Eq. (11): the joint posterior factorization follows Bayes' theorem and Markov assumptions, as in [11].
  • standard math Belief propagation with particle filters from [11] correctly realizes local SLAM
    Section IV-A: the paper reuses the BP-based SLAM solver for components (a) through (c) and does not re-derive or verify it.
  • domain assumption CARLA simulator and YOLOv8 detections are representative of real ISAC sensing
    Section VII: all performance claims are obtained in synthetic indoor and outdoor environments; no real-world validation is provided.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Cooperative Mapping, Localization, and Beam Management via Multi-Modal SLAM in ISAC Systems." pith.science (2026). https://pith.science/paper/WBN4KT2O

@misc{pith2026250705718,
  author       = {Pith},
  title        = {Pith review of: Cooperative Mapping, Localization, and Beam Management via Multi-Modal SLAM in ISAC Systems},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/WBN4KT2O}},
  note         = {Machine review of arXiv:2507.05718}
}
read the original abstract

Simultaneous localization and mapping (SLAM) plays a critical role in integrated sensing and communication (ISAC) systems for sixth-generation (6G) millimeter-wave (mmWave) networks, enabling environmental awareness and precise user equipment (UE) positioning. While cooperative multi-user SLAM has demonstrated potential in leveraging distributed sensing, its application within multi-modal ISAC systems remains limited, particularly in terms of theoretical modeling and communication-layer integration. This paper proposes a novel multi-modal SLAM framework that addresses these limitations through three key contributions. First, a Bayesian estimation framework is developed for cooperative multi-user SLAM, along with a two-stage algorithm for robust radio map construction under dynamic and heterogeneous sensing conditions. Second, a multi-modal localization strategy is introduced, fusing SLAM results with camera-based multi-object tracking and inertial measurement unit (IMU) data via an error-aware model, significantly improving UE localization in multi-user scenarios. Third, a sensing-aided beam management scheme is proposed, utilizing global radio maps and localization data to generate UE-specific prior information for beam selection, thereby reducing inter-user interference and enhancing downlink spectral efficiency. Simulation results demonstrate that the proposed system improves radio map accuracy by up to 60%, enhances localization accuracy by 37.5%, and significantly outperforms traditional methods in both indoor and outdoor environments.

Figures

Figures reproduced from arXiv: 2507.05718 by the authors.

Figure 1
Figure 1. An illustration of the mmWave channel multipath propagation process. [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Block diagram of the proposed cooperative radio map construction, UE localization, and beam management algorithm. [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Multi-user SLAM factor graph. The estimation is categorized into two main areas: local SLAM states and the global radio map state. For the local SLAM state, UEs autonomously determine their positions and map the immediate environment. Conversely, the global radio map state involves the BS or a designated central unit amalgamating the individual SLAM states from all UEs to create a unified global radio map. The local… view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Performance of radio map construction: (a) across different features [PITH_FULL_IMAGE:figures/full_fig_p011_4.png]
Figure 5
Figure 5. Figure 5: Performance of radio map construction in (a) scenario 4, which [PITH_FULL_IMAGE:figures/full_fig_p011_5.png]
Figure 7
Figure 7. Figure 7: (a) the CARLA simulation settings and (b) an illustration of YOLO [PITH_FULL_IMAGE:figures/full_fig_p012_7.png]
Figure 8
Figure 8. Figure 8: Visualization of the look-up table that maps the input pixel coordinates [PITH_FULL_IMAGE:figures/full_fig_p013_8.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

46 extracted references · 46 canonical work pages

  1. [1]

    Bayesian framework for multi-user cooperative radio SLAM,

    H. Que, J. Yang, T. Du, S. Xia, C.-K. Wen, and S. Jin, “Bayesian framework for multi-user cooperative radio SLAM,” in Proc. IEEE 35th Annu. Int. Symp. Pers., Indoor Mobile Radio Commun. (PIMRC) , Valencia, Spain, Sept. 2024, pp. 1–6

    work page 2024

  2. [2]

    Integrated sensing and communications: Toward dual-functional wire- less networks for 6G and beyond,

    F. Liu, Y . Cui, C. Masouros, J. Xu, T. X. Han, Y . C. Eldar, and S. Buzzi, “Integrated sensing and communications: Toward dual-functional wire- less networks for 6G and beyond,” IEEE J. Sel. Areas Commun., vol. 40, no. 6, pp. 1728–1767, Mar. 2022. IEEE TRANSACTIONS, VOL. XX, NO. XX, XX 20XX 14

    work page 2022

  3. [3]

    Integrated sensing and communication in 6G: Motivations, use cases, requirements, challenges and future directions,

    D. K. Pin Tan et al., “Integrated sensing and communication in 6G: Motivations, use cases, requirements, challenges and future directions,” in Proc. 1st IEEE Int. Online Symp. Joint Commun. Sens. (JC&S) , Dresden, Germany, Feb. 2021, pp. 1–6

    work page 2021

  4. [4]

    Simultaneous localization and mapping in multipath environments,

    C. Gentner, B. Ma, M. Ulmschneider, T. Jost, and A. Dammann, “Simultaneous localization and mapping in multipath environments,” in Proc. IEEE/ION Position, Location Navigat. Symp. (PLANS) , Savannah, GA, USA, Apr. 2016, pp. 807–815

    work page 2016

  5. [5]

    Communication-driven localization and mapping for millimeter wave networks,

    J. Palacios et al., “Communication-driven localization and mapping for millimeter wave networks,” in Proc. IEEE Conf. Comput. Commun. (INFOCOM), Honolulu, HI, USA, Apr. 2018, pp. 2402–2410

    work page 2018

  6. [6]

    Location awareness in beyond 5G networks,

    A. Conti, F. Morselli, Z. Liu, S. Bartoletti, S. Mazuelas, W. C. Lindsey, and M. Z. Win, “Location awareness in beyond 5G networks,” IEEE Commun. Mag., vol. 59, no. 11, pp. 22–27, Nov. 2021

    work page 2021

  7. [7]

    Enabling situational awareness in millimeter wave massive MIMO systems,

    R. Mendrzik, F. Meyer, G. Bauch, and M. Z. Win, “Enabling situational awareness in millimeter wave massive MIMO systems,” IEEE J. Sel. Topics Signal Process., vol. 13, no. 5, pp. 1196–1211, Sept. 2019

    work page 2019

  8. [8]

    Neural RF SLAM for unsupervised positioning and mapping with channel state information,

    S. Kadambi, A. Behboodi, J. B. Soriaga, M. Welling, R. Amiri, S. Yer- ramalli, and T. Yoo, “Neural RF SLAM for unsupervised positioning and mapping with channel state information,” in Proc. IEEE Int. Conf. Commun. (ICC), Seoul, Korea, May 2022, pp. 3238–3244

    work page 2022

Show all 46 references
  1. [9]

    5G mmWave cooperative positioning and mapping using multi- model PHD filter and map fusion,

    H. Kim, K. Granstr ¨om, L. Gao, G. Battistelli, S. Kim, and H. Wymeer- sch, “5G mmWave cooperative positioning and mapping using multi- model PHD filter and map fusion,” IEEE Trans. Wireless Commun. , vol. 19, no. 6, pp. 3782–3795, Mar. 2020

    2020

  2. [10]

    MOSAIC: Simultaneous localization and environment mapping using mmWave without a-priori knowledge,

    A.Yassin et al., “MOSAIC: Simultaneous localization and environment mapping using mmWave without a-priori knowledge,” IEEE Access , vol. 6, pp. 68 932–68 947, Nov. 2018

    2018

  3. [11]

    A belief propagation algorithm for multipath-based SLAM,

    E. Leitinger, F. Meyer, F. Hlawatsch, K. Witrisal, F. Tufvesson, and M. Z. Win, “A belief propagation algorithm for multipath-based SLAM,” IEEE Trans. Wireless Commun., vol. 18, no. 12, pp. 5613–5629, Sept. 2019

    2019

  4. [12]

    Angle- based SLAM on 5G mmWave systems: Design, implementation, and measurement,

    J. Yang, C.-K. Wen, J. Xu, H. Que, H. Wei, and S. Jin, “Angle- based SLAM on 5G mmWave systems: Design, implementation, and measurement,” IEEE Internet Things J. , vol. 10, no. 20, pp. 17 755– 17 771, May 2023

    2023

  5. [13]

    Enabling plug-and-play and crowdsourcing SLAM in wireless communication systems,

    J. Yang, C.-K. Wen, S. Jin, and X. Li, “Enabling plug-and-play and crowdsourcing SLAM in wireless communication systems,” IEEE Trans. Wireless Commun., vol. 21, no. 3, pp. 1453–1468, Mar. 2022

    2022

  6. [14]

    Message passing-based 9-D cooperative localization and navigation with embed- ded particle flow,

    L. Wielandner, E. Leitinger, F. Meyer, and K. Witrisal, “Message passing-based 9-D cooperative localization and navigation with embed- ded particle flow,” IEEE Trans. Signal Inf. Process. over Netw. , vol. 9, pp. 95–109, Jan. 2023

    2023

  7. [15]

    Multipath- based SLAM with cooperation and map fusion in MIMO systems,

    E. Leitinger, L. Wielandner, A. Venus, and K. Witrisal, “Multipath- based SLAM with cooperation and map fusion in MIMO systems,”

  8. [16]

    Environment reconstruction based on multi-user selection and multi-modal fusion in ISAC,

    B. Lin, C. Zhao, F. Gao, G. Ye Li, J. Qian, and H. Wang, “Environment reconstruction based on multi-user selection and multi-modal fusion in ISAC,” IEEE Trans. Wireless Commun. , vol. 23, no. 10, pp. 15 083– 15 095, Oct. 2024

    2024

  9. [17]

    Joint vehicular localization and reflective mapping based on team channel- SLAM,

    X. Chu, Z. Lu, D. Gesbert, L. Wang, X. Wen, M. Wu, and M. Li, “Joint vehicular localization and reflective mapping based on team channel- SLAM,” IEEE Trans. Wireless Commun., vol. 21, no. 10, pp. 7957–7974, Apr. 2022

    2022

  10. [18]

    Channel estimation and hybrid precoding for millimeter wave cellular systems,

    A. Alkhateeb et al., “Channel estimation and hybrid precoding for millimeter wave cellular systems,” IEEE J. Sel. Topics Signal Process. , vol. 8, no. 5, pp. 831–846, Jul. 2014

    2014

  11. [19]

    A tutorial on beam management for 3GPP NR at mmwave frequencies,

    M. Giordani, M. Polese, A. Roy, D. Castor, and M. Zorzi, “A tutorial on beam management for 3GPP NR at mmwave frequencies,” IEEE Commun. Surveys Tuts., vol. 21, no. 1, pp. 173–196, Jan. 2019

    2019

  12. [20]

    Detection and tracking of multipath channel parameters using belief propagation,

    X. Li, E. Leitinger, and F. Tufvesson, “Detection and tracking of multipath channel parameters using belief propagation,” in Proc. 54th Asilomar Conf. Signals, Syst., Comput. (ACSSC) , Pacific Grove, CA, USA, Nov. 2020, pp. 1083–1089

    2020

  13. [21]

    Radar-assisted predictive beamforming for vehicular links: Communication served by sensing,

    F. Liu et al., “Radar-assisted predictive beamforming for vehicular links: Communication served by sensing,” IEEE Trans. Wireless Commun. , vol. 19, no. 11, pp. 7704–7719, Nov. 2020

    2020

  14. [22]

    Location-aware predictive beamforming for UA V communications: A deep learning approach,

    C. Liu, W. Yuan, Z. Wei, X. Liu, and D. W. K. Ng, “Location-aware predictive beamforming for UA V communications: A deep learning approach,” IEEE Wireless Commun. Lett. , vol. 10, no. 3, pp. 668–672, Mar. 2021

    2021

  15. [23]

    Joint beam management and SLAM for mmwave communication systems,

    H. Que, J. Yang, C.-K. Wen, S. Xia, X. Li, and S. Jin, “Joint beam management and SLAM for mmwave communication systems,” IEEE Trans. Commun., vol. 71, no. 10, pp. 6162–6179, Jul. 2023

    2023

  16. [24]

    DeepSense 6G: A large-scale real-world multi-modal sensing and communication dataset,

    A. Alkhateeb, G. Charan, T. Osman, A. Hredzak, J. Morais, U. Demirhan, and N. Srinivas, “DeepSense 6G: A large-scale real-world multi-modal sensing and communication dataset,” IEEE Commun. Mag., vol. 61, no. 9, pp. 122–128, Sept. 2023

    2023

  17. [25]

    Position-aided beam prediction in the real world: How useful GPS locations actually are?

    J. Morais et al., “Position-aided beam prediction in the real world: How useful GPS locations actually are?” in Proc. IEEE Int. Conf. Commun. (ICC), Rome, Italy, Jun. 2023, pp. 1824–1829

    2023

  18. [26]

    LiDAR aided future beam prediction in real-world millimeter wave V2I communications,

    S. Jiang, G. Charan, and A. Alkhateeb, “LiDAR aided future beam prediction in real-world millimeter wave V2I communications,” IEEE Wireless Commun. Lett., vol. 12, no. 2, pp. 212–216, Feb. 2023

    2023

  19. [27]

    Radar aided 6G beam prediction: Deep learning algorithms and real-world demonstration,

    U. Demirhan et al., “Radar aided 6G beam prediction: Deep learning algorithms and real-world demonstration,” in IEEE Wireless Commun. Netw. Conf. (WCNC), Austin, TX, USA, Apr. 2022, pp. 2655–2660

    2022

  20. [28]

    Computer vision aided beam tracking in a real-world millimeter wave deployment,

    S. Jiang and A. Alkhateeb, “Computer vision aided beam tracking in a real-world millimeter wave deployment,” in Proc. IEEE Globecom Workshops, Dec. 2022, pp. 142–147

    2022

  21. [29]

    Applying deep-learning-based computer vision to wireless communications: Methodologies, opportunities, and challenges,

    Y . Tian et al., “Applying deep-learning-based computer vision to wireless communications: Methodologies, opportunities, and challenges,” IEEE Open J. Commun. Soc. , vol. 2, pp. 132–143, Dec. 2021

    2021

  22. [30]

    Millimeter wave base stations with cameras: Vision-aided beam and blockage prediction,

    M. Alrabeiah, A. Hredzak, and A. Alkhateeb, “Millimeter wave base stations with cameras: Vision-aided beam and blockage prediction,” in Proc. IEEE 91st Veh. Technol. Conf. (VTC-Spring), May 2020, pp. 1–5

    2020

  23. [31]

    Vision aided beam tracking and frequency handoff for mmwave communications,

    T. Zhang, J. Liu, and F. Gao, “Vision aided beam tracking and frequency handoff for mmwave communications,” in IEEE Conf. on Comput. Commun. Workshops (INFOCOM WKSHPS) , New York, NY , USA, May. 2022, pp. 1–2

    2022

  24. [32]

    Hierarchical codebook- based multiuser beam training for millimeter wave massive MIMO,

    C. Qi, K. Chen, O. A. Dobre, and G. Y . Li, “Hierarchical codebook- based multiuser beam training for millimeter wave massive MIMO,” IEEE Trans. Wireless Commun. , vol. 19, no. 12, pp. 8142–8152, Jul. 2020

    2020

  25. [33]

    Multi-user beam alignment for millimeter wave systems in multi-path environments,

    M. A. Amir Khojastepour, S. Shahsavari, A. Khalili, and E. Erkip, “Multi-user beam alignment for millimeter wave systems in multi-path environments,” in 2020 54th Asilomar Conf. on Signals, Syst., Comput. , CA, USA, Jun. 2020, pp. 549–553

    2020

  26. [34]

    Coordinated beam selection in millimeter wave multi-user MIMO using out-of-band information,

    F. Maschietti et al., “Coordinated beam selection in millimeter wave multi-user MIMO using out-of-band information,” in Proc. IEEE Int. Conf. Commun. (ICC) , Shanghai, China, May 2019, pp. 1–6

    2019

  27. [35]

    Deep cooperation in ISAC system: Resource, node and infrastructure perspectives,

    Z. Wei, H. Liu, Z. Feng, H. Wu, F. Liu, Q. Zhang, and Y . Du, “Deep cooperation in ISAC system: Resource, node and infrastructure perspectives,” IEEE Internet Things Mag. , vol. 7, no. 6, pp. 118–125, Nov. 2024

    2024

  28. [36]

    User identification: A key enabler for multi-user vision-aided communications,

    G. Charan and A. Alkhateeb, “User identification: A key enabler for multi-user vision-aided communications,” IEEE Open J. Veh. Technol. , vol. 5, pp. 472–488, Dec. 2024

    2024

  29. [37]

    Multi-user matching and resource allocation in vision aided communications,

    W. Xu, F. Gao, Y . Zhang, C. Pan, and G. Liu, “Multi-user matching and resource allocation in vision aided communications,” IEEE Trans. Commun., vol. 71, no. 8, pp. 4528–4543, Aug. 2023

    2023

  30. [38]

    Vision-based lane detection and tracking for driver assistance systems: A survey,

    H. Zhou and H. Wang, “Vision-based lane detection and tracking for driver assistance systems: A survey,” in Proc. IEEE Int. Conf. Cybern. Intell. Syst. IEEE Conf. Robot., Automat. Mechatron. , Ningbo, China, 2017, pp. 660–665

    2017

  31. [39]

    A survey on 3D LiDAR localization for autonomous vehicles,

    M. Elhousni and X. Huang, “A survey on 3D LiDAR localization for autonomous vehicles,” in Proc. IEEE Intell. Vehicles Symp. (IV) , Las Vegas, NV , USA, 2020, pp. 1879–1884

    2020

  32. [40]

    Goldsmith, Wireless communications

    A. Goldsmith, Wireless communications. U.K.: Cambridge Univ. Press, 2005

    2005

  33. [41]

    You only look once: Unified, real-time object detection,

    J. Redmon et al., “You only look once: Unified, real-time object detection,” in 2016 IEEE Conf. on Comput. Vis. Pattern Recognit. (CVPR), Las Vegas, NV , USA, Jun. 2016, pp. 779–788

    2016

  34. [42]

    Expandable YOLO: 3D object detection from RGB-D images,

    M. Takahashi, Y . Ji, K. Umeda, and A. Moro, “Expandable YOLO: 3D object detection from RGB-D images,” in 2020 21st Int. Conf. on Res. Educ. Mechatronics (REM) , Cracow, Poland, Jan. 2020, pp. 1–5

    2020

  35. [43]

    Multipath assisted positioning with simultaneous localization and mapping,

    C. Gentner, T. Jost, W. Wang, S. Zhang, A. Dammann, and U.-C. Fiebig, “Multipath assisted positioning with simultaneous localization and mapping,” IEEE Trans. Wireless Commun., vol. 15, no. 9, pp. 6104– 6117, Sept. 2016

    2016

  36. [44]

    CARLA: An open urban driving simulator,

    A. Dosovitskiy, G. Ros, F. Codevilla, A. Lopez, and V . Koltun, “CARLA: An open urban driving simulator,” in Proc. 1st Annu. Conf. Robot Learn. (PMLR) , Nov. 2017, pp. 1–16

    2017

  37. [45]

    YOLO by ultralytics,

    G. Jocher, A. Chaurasia, and J. Qiu, “YOLO by ultralytics,” https:// github.com/ultralytics/ultralytics, Jan. 2023

    2023

  38. [2024]

    Available: https://arxiv.org/abs/2405.02126

    [Online]. Available: https://arxiv.org/abs/2405.02126

    arXiv

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.