{"id":"3c2c6b93-a40f-4b27-a25f-019f3d227914","arxiv_id":"2506.09225","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":3.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Near-field sensing of angle, distance, radial velocity, and transverse velocity enables trajectory-model-free predictive beamforming for high-mobility wireless users.","lead":"The paper proposes a near-field predictive beamforming framework that uses spherical waves and non-uniform Doppler shifts to estimate a user's full position and velocity, then predicts future positions for proactive beam steering without trajectory models. A generalist might read it because it points toward simpler and more general beam steering for high-mobility 6G networks such as vehicles and drones.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Kinematic update in Sec. III-A is dimensionally inconsistent and imposes a constant-polar-velocity prior, undermining the 'arbitrary trajectory / prior-knowledge-free' claim.","rationale":"The paper's central claim is that near-field sensing enables predictive beamforming without prior trajectory knowledge. The entire prediction block rests on the two-line kinematic model in Sec. III-A. That model contains a unit inconsistency: θ' = θ + vθΔT. The second term has units of length, yet θ' is an angle. The correct polar kinematic update for constant polar-velocity motion is θ' = θ + (vθ ΔT)/r. If the authors intended something else, they need to define it; as written, the model is not a valid update rule. Second, holding vr and vθ constant is a stronger assumption than ordinary constant-velocity motion; a straight-line uniformly moving target has changing radial/transverse components. Therefore the claim of handling 'arbitrary trajectories' is not supported by the evidence. The reader's weakest assumption concerned the reliability of estimation under multipath/noise. That is also valid, but I view the kinematic model error as more load-bearing because it holds even in the idealized noiseless case. The case study in Sec. III-C shows only qualitative plots and no numerical error metrics, so the central advantages are not quantitatively validated. My recommended verdict is unchanged: the paper should be conditional pending correction of the kinematic model and a quantitative evaluation, ideally including a non-cooperative (maneuvering) trajectory. I partially disagree with the reader's pinpointed weak assumption: both issues matter, but the model inconsistency is prior to the estimation problem.","tokens_in":9942,"tokens_out":4933,"duration_ms":53342,"concrete_test":"Re-derive the kinematic update in Section III-A with unit consistency and re-run the case study with (i) the corrected polar update θ' = θ + vθΔT/r and (ii) a ground-truth trajectory that is a straight line with constant Cartesian velocity. Feed perfect noise-free measurements of (r, θ, vr, vθ) into the model. If the predicted (r', θ') diverges from the true next-CPI position for the straight-line target, the framework does not generalize to arbitrary trajectories as claimed.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section III-A's prediction step uses r' = r + vr ΔT and θ' = θ + vθ ΔT. Since r is in meters and θ is an angle in radians, the second equation has inconsistent units (vθ ΔT gives meters, not radians). The correct polar-coordinate update is θ' = θ + (vθ ΔT)/r (or an equivalent Cartesian update). As written, the model is incorrect and cannot be the basis of the framework. Moreover, even with the corrected equation, holding vr and vθ constant across CPIs is a 'constant polar velocity' model, not a general constant-velocity model: a target moving in a straight line with constant Cartesian velocity has time-varying vr and vθ. Thus the claimed support for 'arbitrary trajectories' and 'prior-knowledge-free prediction' is not established by the case study, which never reports quantitative prediction error or uses a maneuvering trajectory. The dimension error is internal to the paper, not a matter of consensus.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes a near-field predictive beamforming (NFPB) framework for high-mobility wireless networks. It argues that the spherical wavefronts and non-uniform Doppler shifts in the near-field enable, for the first time, joint estimation of angle, distance, radial velocity, and transverse velocity, which in turn allows predictive beamforming without a predefined trajectory state-evolution model. The paper reviews the relevant near-field sensing fundamentals, presents a five-step NFPB framework with a kinematic prediction model, discusses estimation-, filtering-, and learning-based implementation approaches, and provides a simulation-style case study with qualitative tracking plots. It then lists several research opportunities (industrial IoT, V2X, high-speed scenarios, RIS, flexible antennas, UAV, and physical-layer security).","tokens_in":10104,"tokens_out":2318,"duration_ms":24264,"significance":"If the central claims were established, the framework would be a meaningful conceptual advance: it would remove the need for trajectory-specific state evolution models in predictive beamforming and eliminate dedicated feedback/pilot overhead, with potential impact on 6G ISAC design. The paper usefully synthesizes the physical rationale for near-field full-dimensional sensing and articulates a clear comparison with far-field sensing. However, the paper's own validation is limited to a qualitative case study, and the kinematic model contains a dimensional inconsistency that undermines the framework's technical foundation. The significance is therefore conditional on substantial revision and quantitative validation.","major_comments":[{"comment":"The kinematic update for the polar angle is dimensionally inconsistent: the equation θ′ = θ + vθ ΔT adds a quantity with units of meters (vθ in m/s times ΔT in s) to an angle in radians. The correct polar-coordinate update is θ′ = θ + (vθ ΔT)/r, or equivalently an update performed in Cartesian coordinates. Since this model is the basis for the prediction step of the entire NFPB framework, the error is load-bearing and must be corrected.","section":"Section III-A, Eq. (1)"},{"comment":"The claim of 'Prior-Knowledge-Free Prediction' is overstated. The kinematic model assumes that vr and vθ remain constant across consecutive CPIs, which is itself a specific prior (a constant-polar-velocity model). A target moving with constant Cartesian velocity generally has time-varying radial and transverse velocities, so the framework does not support 'arbitrary trajectories' as claimed. The case study in Section III-C does not exercise a maneuvering trajectory, so it provides no evidence that the framework generalizes beyond its implicit constant-velocity prior.","section":"Section III-A"},{"comment":"The case study presents only qualitative plots (Figs. 3 and 4) with no quantitative tracking-error metrics (e.g., RMSE or bias), no comparison with a far-field baseline or with a state-evolution-model-based method, and no evaluation of the resulting communication performance (e.g., achievable rate or beamforming gain). Additionally, the beamforming vector used in the case study is never formally defined or derived. These omissions mean that the claimed benefits of NFPB—prior-knowledge-free prediction, low complexity, and generalizability—are not substantiated by the presented results.","section":"Section III-C"}],"minor_comments":[{"comment":"The heading 'NFPB for V2X Netwroks' contains a typo; it should read 'Networks'.","section":"Section IV-B heading"},{"comment":"The phrase 'frequency shits' should be 'frequency shifts'.","section":"Section III-A, step 4"},{"comment":"References [9] and [10] are the authors' own prior works and are central to the implementation methods; the paper should clarify, for each, the extent to which the case study re-uses those methods versus providing new results.","section":"References"}],"recommendation":"major_revision","confidential_remarks":"The paper is largely a framework/overview article whose technical core is drawn from the authors' prior work ([9], [10]). The dimensional inconsistency in the kinematic model and the absence of quantitative validation are important, but both are fixable within the manuscript's scope. The fit with a magazine-style venue is reasonable if the claims are tightened. I would also encourage the editor to ensure that the self-citation pattern is not masking a lack of independent validation."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear colleague,\n\nThe key thing to know about this paper is that its central predictive equation is dimensionally wrong. In Section III-A, the polar-coordinate update is given as r' = r + vr ΔT and θ' = θ + vθ ΔT. The second line adds meters (vθ ΔT) to an angle. The correct update is θ' = θ + (vθ ΔT)/r. This isn't a cosmetic slip: the prediction step is the mechanism that supposedly delivers prior-knowledge-free beamforming, and as written it cannot work.\n\nThat said, the paper has some real strengths. The exposition of near-field full-dimensional sensing is clear: spherical waves give joint angle-distance estimation, and non-uniform Doppler gives radial and transverse velocity estimates. The RCRB plots in Fig. 1 nicely illustrate the smooth transition from near-field to far-field behavior. The authors also correctly identify the core problem with far-field predictive beamforming—radial velocity alone cannot update the angle without a state-evolution model.\n\nThe soft spots beyond the equation error are the usual ones for this genre. The case study is an illustrative re-run of the authors' own Kalman-filter method from [9], [10]; there are no tracking-error numbers, no comparison to a far-field baseline, and no beamforming gain curves. The claim that the framework handles \"arbitrary trajectories\" is asserted rather than demonstrated, and even after fixing the polar update, holding vr and vθ constant across CPIs is a constant-polar-velocity assumption, not a general constant-velocity model. The paper also leans heavily on the authors' prior work; for a reader familiar with those papers, the new contribution is mostly a framework narrative.\n\nIf you're writing a tutorial or a survey on near-field ISAC, this is a handy reference for the fundamentals. As a research contribution, however, the dimensional error undercuts the main claim. I'd recommend not accepting it in current form. A serious referee should flag the equation, and the authors would need to redo the case study with quantitative metrics. I wouldn't cite it as-is, but I'd keep an eye on a revised version.\n\nBest,\n[name]","headline":"The kinematic update equation in Section III-A is dimensionally inconsistent, and the case study lacks the quantitative evidence needed to support the framework's claims.","tokens_in":10637,"tokens_out":4673,"would_cite":false,"duration_ms":42432,"reading_group":"maybe","serious_thinker":"no","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Near-field sensing can deliver angle, distance, radial velocity, and transverse velocity from a single echo, letting predictive beamforming track arbitrary user trajectories without any state evolution model.","keywords":["near-field communications","predictive beamforming","integrated sensing and communication","extremely large antenna arrays","non-uniform Doppler","full-dimensional sensing","high-mobility networks","kinematic tracking"],"falsifier":"Run the NFPB loop with a 256-element array at 30 GHz, a target near the Rayleigh distance moving transversely at 20 m/s, and SNR at 0 dB; estimate angle, distance, radial velocity, and transverse velocity from the echo, propagate one coherent processing interval, and check whether the predicted angle error stays below the beam's half-power beamwidth. If it misses more than a few percent of the time in a multipath-rich environment, the prior-knowledge-free claim fails in that regime.","tokens_in":1725,"feed_emoji":"📡","tokens_out":2083,"duration_ms":93807,"temperature":0.7,"pith_summary":"The paper argues that in the near-field region, where antenna arrays are large enough that wavefront curvature and Doppler shifts vary across the aperture, a base station can recover a user's full mobility state: angle, distance, radial velocity, and transverse velocity, all from reflected echoes. With all four quantities available, the transmitter can update the user's position with a short-time kinematic model and steer the next beam proactively, without needing any scenario-specific trajectory model. That removes the main limitation of far-field predictive beamforming, which can measure only angle and radial velocity and therefore must guess future motion from a state evolution model. If the claim holds, high-mobility links in dense urban, vehicular, and aerial settings can keep beams locked on users while skipping pilot-based channel estimation and dedicated uplink feedback.","feed_headline":"Near-field sensing enables model-free predictive beamforming","feed_subtitle":"Spherical waves reveal all motion parameters, so beamforming needs no trajectory model or pilot overhead.","key_machinery":"The mechanism that carries the argument is the near-field phase structure across an extremely large antenna array. Spherical wavefronts arriving from the same angle but different distances have different curvatures, so distance becomes observable alongside angle; non-uniform Doppler shifts across the aperture make transverse velocity observable alongside radial velocity. The paper formalizes this transition through the root Cramer-Rao bound (RCRB): as target distance grows, the RCRB for distance and transverse velocity rises while the RCRB for angle and radial velocity falls, showing a smooth degeneration into far-field sensing. This phase structure feeds the NFPB loop: echo signal, full-dimensional mobility estimate, kinematic prediction, array gain maximization with Doppler compensation, and the resulting predictive beamforming vector.","core_discovery":"The central discovery offered is that near-field propagation turns predictive beamforming into a fully observable prediction problem rather than a model-dependent extrapolation problem. Because spherical waves make phase curvature distance-dependent, angle and distance can be estimated jointly; because Doppler shifts are non-uniform across the array, radial and transverse velocity can be estimated jointly. Feeding these four estimates into the paper's kinematic update, $r' = r + v_r \\Delta T$ and $\\theta' = \\theta + v_\\theta \\Delta T$, yields the predicted polar position for the next coherent processing interval, and the beamforming vector is designed to focus at that predicted location while compensating the predicted Doppler. The paper calls the resulting framework near-field predictive beamforming (NFPB) and claims two advantages over far-field predictive beamforming: prediction without prior trajectory knowledge, and system design that is lower complexity and re-usable when scenarios change. The included case study with a Kalman-filter-based tracker shows accurate position and velocity tracking of a trajectory with changing radius, without any state evolution model.","pith_inferences":["A consequence the authors leave implicit: NFPB's model-free advantage depends on a resolvable line-of-sight echo, so in rich scattering the four-parameter estimation can degrade; their listed RIS-aided direction is one way to restore that condition.","Because the framework measures transverse velocity directly, it effectively yields angular-rate information; fusing that with map or inertial data during brief non-line-of-sight outages is a natural extension the paper does not develop.","All four mobility parameters are estimated per coherent processing interval, so neighboring base stations could exchange these state estimates for cooperative scheduling; the paper mentions coordination in V2X networks but does not specify a protocol.","A direct test: compare NFPB's achieved beam gain against far-field predictive beamforming that is given a perfectly known state evolution model; on arbitrary trajectories NFPB should match or exceed it while avoiding the overhead of model derivation."],"forward_implications":["Near-field predictive beamforming can eliminate dedicated pilot overhead for channel estimation and dedicated uplink feedback, freeing coherent time for data transmission.","The framework applies to users following arbitrary trajectories because it needs no pre-derived state evolution model; urban V2I, UAV, and industrial IoT links are named as direct beneficiaries.","Because the full mobility state is measured at each coherent processing interval, the same system design carries over when the scenario changes, reducing redesign effort.","For three-dimensional motion such as UAVs, spherical waves in 3D space provide azimuth, elevation, and distance, extending the same principle to 3D localization and velocity sensing.","With accurate full-dimensional tracking, the beam can follow the intended user tightly, which the paper argues strengthens physical-layer security against eavesdropping and jamming."],"supporting_citations":[{"why":"Defines conventional far-field radar-assisted predictive beamforming and its reliance on a state evolution model, the baseline NFPB improves on.","marker":"[3]"},{"why":"Establishes that the Rayleigh distance can extend to tens or hundreds of meters, making the near-field regime relevant for extremely large antenna array deployments.","marker":"[6]"},{"why":"Supports the claim that spherical-wave phase curvature enables joint angle and distance estimation in near-field MIMO.","marker":"[7]"},{"why":"Derives performance bounds showing non-uniform Doppler across large apertures makes transverse velocity observable, underpinning full-dimensional velocity sensing.","marker":"[8]"},{"why":"Supplies the near-field velocity-projection method and the kinematic-model prediction step NFPB adopts.","marker":"[9]"},{"why":"Provides the filter-based tracking approach from estimation to tracking that the case study uses to demonstrate robust mobility prediction.","marker":"[10]"}],"fun_headline_variants":["Near-field sensing makes predictive beamforming model-free","Model-free predictive beamforming via near-field sensing","Near-field sensing enables prediction without trajectory models","Spherical waves unlock full parameter sensing for beamforming"],"cache_read_input_tokens":12800,"weakest_assumption_plain":"The framework presupposes that a single line-of-sight echo supports reliable joint estimation of angle, distance, radial velocity, and transverse velocity within one coherent processing interval; if multipath, noise, or model mismatch degrades any of these estimates, the kinematic prediction and the beamforming design lose accuracy, and the paper's own case study does not quantify these estimation errors.","fun_headline_variants_meta":{"raw":{"variants":["Near-field sensing makes predictive beamforming model-free","Model-free predictive beamforming via near-field sensing","Near-field sensing enables prediction without trajectory models","Spherical waves unlock full parameter sensing for beamforming"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000362,"raw_usage":{"total_tokens":1918,"prompt_tokens":874,"completion_tokens":1044,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":490,"completion_tokens_details":{"reasoning_tokens":985}},"tokens_in":490,"tokens_out":1044,"duration_ms":7912,"temperature":1.0,"reasoning_tokens":985,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T04:53:04.737850+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the NFPB loop with a 256-element array at 30 GHz, a target near the Rayleigh distance moving transversely at 20 m/s, and SNR at 0 dB; estimate angle, distance, radial velocity, and transverse velocity from the echo, propagate one coherent processing interval, and check whether the predicted angle error stays below the beam's half-power beamwidth. If it misses more than a few percent of the time in a multipath-rich environment, the prior-knowledge-free claim fails in that regime.","supporting_citations":[{"cited_title":"Radar-assisted predictive beamforming for vehicular li nks: Communication served by sensing,","cited_arxiv_id":null,"evidence_quote":"Defines conventional far-field radar-assisted predictive beamforming and its reliance on a state evolution model, the baseline NFPB improves on."},{"cited_title":"Hierarchical beam training for extremely large-scale MIMO: From far-field to near-field,","cited_arxiv_id":null,"evidence_quote":"Establishes that the Rayleigh distance can extend to tens or hundreds of meters, making the near-field regime relevant for extremely large antenna array deployments."},{"cited_title":"Near-field MIMO communications for 6G: Fundamentals, challenges, potentials, and future directions,","cited_arxiv_id":null,"evidence_quote":"Supports the claim that spherical-wave phase curvature enables joint angle and distance estimation in near-field MIMO."},{"cited_title":"Performance bounds for velocity estimation with extremely large aperture arrays,","cited_arxiv_id":null,"evidence_quote":"Derives performance bounds showing non-uniform Doppler across large apertures makes transverse velocity observable, underpinning full-dimensional velocity sensing."},{"cited_title":"Near-field velocity sensing and predictive beamform- ing,","cited_arxiv_id":null,"evidence_quote":"Supplies the near-field velocity-projection method and the kinematic-model prediction step NFPB adopts."}],"review_version":1}