Multiple Access in Pinching-Antenna Systems: A Survey, Impairment Analysis, and Evaluation

I. INTRODUCTION

• PASS can place pinches near users along a long, low-loss waveguide, while pinches on one guide share an RF chain and radiate the same composite signal.This makes multiple access fundamentally different from a co-located array with a dedicated precoded signal at each element.Source: §I · p. 1
• A single waveguide has no spatial-domain access: users must be separated in time, frequency, power, or guided mode.The shared feed also dilutes transmit power across active pinches, imposes a multi-user phase-alignment conflict, and causes uplink inter-antenna radiation.Source: §I · p. 1

A. State of the Art

• The survey organizes PASS access research by the waveguide-to-user ratio, distinguishing the single-waveguide regime from multi-waveguide and cross-regime families.A single guide is the hardest-constrained building block; multiple guides add spatial access.Source: A. State of the Art · p. 2

1) Single-Waveguide Families:

• Single-waveguide TDMA can be optimal or outperform NOMA in several settings, while FDMA and OFDMA separate users by sub-band or subcarrier.TDMA results depend on activation and placement: per-slot re-pinching achieves uplink sum capacity in one setting, and OFDMA accommodates frequency selectivity with a cyclic prefix.Source: 1) Single-Waveguide Families · p. 3
• Power-domain NOMA dominates single-waveguide research, with pinch placement used to engineer the channel-gain disparity required for SIC.The surveyed work covers placement, power allocation, rate and error analysis, uplink, fairness, robustness, security, and applications,, –[36].Source: 1) Single-Waveguide Families · p. 3
• Mode-division multiplexing uses multiple guided modes and RF chains, while segmented-waveguide designs use segment selection, aggregation, or multiplexing.Segmentation removes uplink IAR with one active pinch per segment; among its modes, SM>SA>SS, and SA captures most of SM's gain on one RF chain [39].Source: 1) Single-Waveguide Families · p. 3

2) Multi-Waveguide Families:

• Waveguide multiplexing uses baseband precoding across guides to separate users in space, while waveguide- and environment-division access use guides or propagation segments without joint baseband processing.Power-domain NOMA can also be overlaid within each guide.Source: 2) Multi-Waveguide Families · p. 3
• Multi-cell PASS work addresses inter-cell interference through joint transmission, cell-free coordination, geometry, or deployment analysis.The surveyed approaches include coherent transmission, pilot-contamination mitigation, and geometry-based interference reduction.Source: 2) Multi-Waveguide Families · p. 3

3) Cross-Regime and Emerging Families:

• Rate splitting spans single- and multi-waveguide and uplink and downlink settings, splitting messages into common and private parts to interpolate between SDMA and NOMA,.Index and spatial modulation encode bits through the radiating pinch or its position, but surveyed schemes are point-to-point,.Source: 3) Cross-Regime and Emerging Families · p. 5
• Code-domain access has no known PASS instantiation, while grant-free or massive random access has only a single broached SWAN-based contention protocol.The cited survey identifies these as open areas in the access-family landscape.Source: 3) Cross-Regime and Emerging Families · p. 5

B. The Idealization Gap, Scope, and Contributions

• Of 48 surveyed access-family works, 37 model none of the article's in-scope impairments and the other 11 model at most two, each in isolation.The prevailing idealizations include pure LoS, perfect CSI and SIC, negligible IWA, instantaneous repositioning, and omnidirectional radiation; omnioptimized placements lose about 7.5 dB [58].Source: B. The Idealization Gap, Scope, and… · p. 5
• The article combines a spatial-DoF taxonomy, an impairment-versus-method analysis, and a unified benchmark of time-, frequency-, power-, and space-domain access.Its cross-cutting finding is that schemes are generally most vulnerable when an impairment corrupts their user-separation resource; it reports no universal scheme ranking.Source: B. The Idealization Gap, Scope, and… · p. 5

C. Notation and Organization

• The article proceeds from notation and a system model to impairment analysis and then a unified benchmark.The system model covers power allocation across pinches, downlink and uplink channels, multi-waveguide architectures, and shared-feed penalties.Source: C. Notation and Organization · p. 6, §II · p. 6

II. SYSTEM MODEL

• PASS pinches couple guided power to free space, and their positions can be added, removed, or repositioned to configure the radiating geometry.The model distinguishes equal-power pinches from identical pinches that radiate a proportional share of the remaining guided power.Source: §II · p. 6
• Equal-power and proportional-power conventions nearly coincide for short, low-loss guides with few pinches, but diverge as pinch count and in-waveguide attenuation grow.The distinction matters for large active sets, long guides, and terahertz operation.Source: §II · p. 6

A. Downlink Signal Model

• Each pinch contributes a received signal shaped by its free-space distance to the user and its guided-wave distance from the feed.Moving one pinch changes its amplitude and both free-space and in-waveguide phases; the guided-wavelength-scale oscillations create dense local optima and sensitivity to position error.Source: A. Downlink Signal Model · p. 7
• Splitting the power budget among pinches can make array gain non-monotone, while one shared phase profile generally cannot align signals constructively at multiple users.The phase conflict arises when at least two pinches serve at least two users.Source: A. Downlink Signal Model · p. 7

B. Uplink Signal Model

• Inter-antenna radiation couples active pinches on the shared guide as the user’s returning signal travels toward the feed.It is the uplink counterpart of downlink power dilution.Source: B. Uplink Signal Model · p. 7
• Prior uplink models either neglect pinch re-radiation for tractability or activate one PA per slot or segment to avoid it.The paper calls the latter models IAR-free.Source: B. Uplink Signal Model · p. 7
• The paper’s IAR-aware model carries the coupled-mode description to the uplink, depleting the returning wave at each pinch it passes.The depletion is described as first order.Source: B. Uplink Signal Model · p. 7
• The paper relates its uplink model to prior approaches: removing the cascade recovers IAR-neglecting transfer, while one active PA makes it exact.The single-active-PA case corresponds to IAR-free operation.Source: B. Uplink Signal Model · p. 7

C. Multi-Waveguide Signal Model

• Independent streams across multiple waveguides enable spatial multiplexing, which is impossible with one guide’s single composite signal.With M ≥ 2 guides, spatial separation is available; one stream per user is possible once M ≥ K.Source: C. Multi-Waveguide Signal Model · p. 8
• Waveguide switching, division, and multiplexing trade off guide use and baseband processing, with full precoding providing genuine SDMA.For 1 < M < K, clustered hybrids can run NOMA within each guide; inter-waveguide coupling and multi-chain synchronization are costs.Source: C. Multi-Waveguide Signal Model · p. 8

III. ANALYSIS OF THE MULTIPLE-ACCESS–CRITICAL IMPAIRMENTS

• The impairment analysis retains effects that reshape multiple-access rates through channel perturbation, orthogonality leakage, or structural coupling on the shared-guide uplink.It presents the resulting severity in an impairment-versus-method map.Source: §III · p. 8

A. Unified Analytical Framework

• A unified rate form exposes how realistic impairments affect access schemes through channel changes or residual inter-user interference.The framework represents these mechanisms as ∆g and ε, respectively.Source: A. Unified Analytical Framework · pp. 8–9
• Channel perturbations matter to multiple access according to how much user separation depends on the channel; power-domain access is especially exposed because it uses channel magnitude for power splits and SIC order.The framework distinguishes these from orthogonality leakage, which affects the resource used to separate users.Source: A. Unified Analytical Framework · p. 9
• The analysis carries forward per-domain leakage and PASS-specific, decision-changing channel perturbations, while excluding reliability effects and generic impairments treated as conventional-array link effects.Examples named as out of scope include actuator wear and impedance-mismatch reflections.Source: A. Unified Analytical Framework · p. 9

Channel-Perturbing Impairments (∆g): Position Error and the Directional Pattern

• At 28 GHz, a 1 mm position shift rotates phase by about 47° and can move BER by ∼10 dB.Position error disrupts coherent combining, with mean power falling to the incoherent sum for independent phase errors; a single pinch loses no rate. In NOMA, phase mismatch leaves residual cancellation interference that can impose a high-power rate ceiling.Source: 1) Channel-Perturbing Impairments… · pp. 9–10
• Directional pinch patterns can cost several decibels of effective gain when placements assume omnidirectional radiation [58], and can disrupt NOMA power splits and SDMA nulls [58],.Orthogonal access is comparatively forgiving because a channel error for a user served alone does not spill into another user's signal.Source: 1) Channel-Perturbing Impairments… · p. 10

Reconfiguration Latency and Jitter (ε, time) →TDMA

• For a 10% rate gain, re-pinching may take at most about 9% of a slot, or 0.45 ms for a 10 ms frame shared by two users.Latency reduces data time, while jitter requires a longer guard time; continuous mechanical re-pinching can consume an entire slot or more for delay-sensitive traffic, leaving no time for data.Source: 2) Reconfiguration Latency and Jitter… · pp. 10–11

Dispersion and Beam Squint (ε, frequency) → FDMA/OFDMA

• With neff = 1.4 and S = 0.6 m, στ ≈ 0.81 ns and Bc ≈ 200 MHz; doubling the span halves Bc.Geometric delay differences create beam squint and frequency selectivity, reducing coherent signal gain away from the carrier; dispersion and frequency-dependent coupling add further bandwidth-dependent effects.Source: 3) Dispersion and Beam Squint (ε,… · p. 11
• OFDMA avoids delay-induced interference when subcarriers are much narrower than Bc and the cyclic prefix covers the delay differences, whereas one-tap single-carrier equalization can cause a power-independent rate ceiling.An ideal equalizer that handles all channel taps avoids this leakage, leaving only the signal loss from beam squint.Source: 3) Dispersion and Beam Squint (ε,… · p. 11

Crosstalk and Synchronization Errors (ε, space and mode) →SDMA and MDM

• Crosstalk and RF-chain phase or timing misalignment invalidate nominal SDMA precoder nulls, producing a rate ceiling that tends to fall as more streams are served.Signal and interference both grow with transmit power, so increasing power cannot remove the leakage.Source: 4) Crosstalk and Synchronization… · p. 12
• Waveguide division is immune to RF-chain phase misalignment, but crosstalk leaks streams between guides; mode division also suffers from nonideal mode orthogonality and mode coupling.Waveguide-division rate can saturate even without errors because of leakage through the air, while mode coupling caps the rate as in the spatial case.Source: 4) Crosstalk and Synchronization… · p. 12

Imperfect SIC and SIC-Order Errors (ε, power) → NOMA

• Imperfect SIC leaves residual interference and a high-SNR error floor [28],; if impairments invert the SIC order, the strong user may fail to decode the weak user's message.The rate ceiling worsens as the mismatch between known and true channels grows; a phase-only mismatch leaves the SIC order intact.Source: 5) Imperfect SIC and SIC-Order Errors… · pp. 12–13

• General SIC ordering remains an open design problem: dynamic ordering reduces exhaustive search, conservative SINR evaluation keeps SIC feasible under channel uncertainty [32], and multi-guide optimal order follows interference-plus-noise-to-channel-gain ratio.Source: 5) Imperfect SIC and SIC-Order Errors… · p. 13

6) Inter-Antenna Radiation →Shared-Waveguide Uplink

• IAR affects shared-waveguide uplink schemes with multiple active pinches; one active pinch per slot or segment avoids it, while segmentation requires a feed and wired link per segment,.Source: 6) Inter-Antenna Radiation… · p. 13
• With identical pinches, IAR makes received power fall geometrically with each pinch passed, and NOMA interference powers carry the same attenuation factors.Source: 6) Inter-Antenna Radiation… · p. 13
• Because attenuation depends on serving-pinch location, IAR can reorder powers at the feed, disrupting the power split and SIC order on which NOMA relies.Source: 6) Inter-Antenna Radiation… · p. 13
• The first-order IAR factor omits interactions through radiated waves, and a complete end-to-end uplink model remains open.Source: 6) Inter-Antenna Radiation… · p. 13

C. Summary of Impacts and Severity Map

• Interference-raising impairments are most harmful because they impose a rate ceiling that more power cannot overcome.Position error, radiation pattern, beam squint, and inter-antenna radiation lower the signal term; latency and timing jitter reduce available time, while cancellation residuals, delayed copies, crosstalk, and misalignment increase interference.Source: C. Summary of Impacts and Severity Map · p. 13
• An impairment generally harms a scheme most when it corrupts the domain that scheme uses to separate users.The affected mechanisms include per-slot re-pinching, spatial nulls, orthogonal modes, and cancellation; schemes not using the corrupted mechanism are affected less or not at all.Source: C. Summary of Impacts and Severity Map · p. 13
• Table V marks each leakage as critical for access in its own domain, while channel perturbations are at least moderately severe for every scheme.Channel perturbations are critical for NOMA and SDMA, where they also raise interference; rate splitting is milder than NOMA because it performs only one cancellation. Under position error, single-pinch orthogonal access is most robust, with unchanged rate.Source: C. Summary of Impacts and Severity Map · pp. 13–14

IV. UNIFIED IMPAIRMENT-AWARE BENCHMARK

• The unified benchmark compares time-, frequency-, power-, and space-domain access under identical channel and impairment models, and evaluates uplink inter-antenna radiation to first order.Each impairment is switched on alone, and schemes share user placements and impairment draws; mode division, environment division, segmented waveguides, and multi-cell operation are excluded because they need further models.Source: §IV · p. 14

A. Benchmark Setup

• The benchmark serves two stationary users in a 10 m square, with one guide at 3 m height and two guides at y = ±D/3 for space-domain access.The minimum user rate is the worse-served user’s rate; each scheme optimizes its resource allocation to maximize this metric within the power budget.Source: A. Benchmark Setup · p. 14
• For low-loss guides, neglecting attenuation changes minimum user rates by less than 0.1 bit/s/Hz and leaves scheme rankings unchanged.The benchmark excludes lossier guides, with attenuation up to 1.3 dB/m.Source: A. Benchmark Setup · p. 14
• The reported rate spans 8.8 to 15.5 bit/s/Hz between 10 and 30 dBm, and the passage says this reported range matches.Fig. 4 plots minimum user rate versus pinch count without impairments at P = 20 dBm; the solid line is from (4), and markers denote simulation.Source: A. Benchmark Setup · p. 14, Figure 4 · p. 14

B. Baseline Without Impairments

• 0.5 bit/s/Hz per doubling: per-slot re-pinching’s minimum user rate grows by this amount as pinch count increases.Each doubling of N doubles the served user’s SNR and adds 1 bit/s/Hz in its slot; each user holds the guide for about half the frame.Source: B. Baseline Without Impairments · p. 14
• 0.15 to 0.22 bit/s/Hz: fixed-placement time division falls below per-slot re-pinching.This comparison uses the best fixed placement; all quoted differences are paired, with 95% confidence intervals below ±0.01 bit/s/Hz.Source: B. Baseline Without Impairments · p. 15
• 0.15 to 0.17 bit/s/Hz: fixed-placement NOMA leads per-slot re-pinching in 96% of user placements at N = 2.The lead requires high SNR and vanishes for users at equal distance from the guide; below a minimum user rate of about 2 to 3 bit/s/Hz, per-slot re-pinching is ahead.Source: B. Baseline Without Impairments · p. 15

C. Position Error

• One-active-pinch TDMA keeps its rate at every position error, whereas two-pinch TDMA loses less than 0.2% through 0.2 mm and 5% at 1 mm.The two pinches are fine-positioned within ±5 cm without error; without error, two-pinch TDMA reaches 6.35 bit/s/Hz and NOMA 6.45 bit/s/Hz.Source: C. Position Error · p. 15
• At 0.1 mm, raising NOMA power from 20 to 30 dBm lifts its minimum user rate by only 0.14 bit/s/Hz, versus 1.66 bit/s/Hz without error; rate splitting stays within 0.13 bit/s/Hz of two-pinch TDMA through 0.5 mm.Beyond about 0.01 mm, rate splitting falls below TDMA because its common stream is limited by the weaker user; re-estimating the channel after placement lets NOMA lead by about 0.10 bit/s/Hz up to 0.2 mm.Source: C. Position Error · p. 15
• NOMA falls below two-pinch TDMA beyond about 0.007 mm error and below one-active-pinch TDMA beyond about 0.017 mm, as cancellation residuals enter the strong user’s interference.An error of about one degree, reached at 0.02 mm, already leaves a cancellation residual above the noise.Source: C. Position Error · p. 15
• At 0.2 mm, with at least 1 bit/s/Hz required per user, NOMA reaches 8.8 bit/s/Hz sum rate, rate splitting 12.5 bit/s/Hz, and two-pinch TDMA 12.9 bit/s/Hz.Without the per-user rate condition, every two-pinch scheme serves only the stronger user and reaches the same sum rate.Source: C. Position Error · p. 15

D. Reconfiguration Latency

• Guard time scales the rate of every scheme that reconfigures per slot by factor η without changing its SINR; schemes with fixed placement avoid this rate penalty.The rate curves follow the no-guard-time rates, with time shares reoptimized for each guard time.Source: D. Reconfiguration Latency · p. 15
• A 0.11 ms guard time per slot consumes per-slot re-pinching’s 2.3% lead over the best fixed placement.The guard-time budget falls from 0.15 ms at 10 dBm to 0.09 ms at 30 dBm; a longer frame widens it proportionally.Source: D. Reconfiguration Latency · p. 15
• NOMA with its best fixed placement stays ahead of per-slot re-pinching at every guard time.The 0.11 ms budget lies within the range of discrete switching and below that of continuous actuation; even for a 1 s frame, the 11 ms budget is shorter than a 20–30 ms mechanical move.Source: D. Reconfiguration Latency · p. 15

E. Beam Squint

• At 1 GHz, the one-tap receiver’s minimum user rate falls to 0.89 bit/s/Hz, while OFDMA loses 5% and NOMA with OFDM loses 11%.The one-tap rate halves at about 40 MHz and approaches a 2.10 bit/s/Hz ceiling at 100 MHz; the ideal decision-feedback equalizer loses 12% at 1 GHz, while per-slot re-pinching keeps its rate.Source: E. Beam Squint · pp. 15–16, Figure 7 · p. 17

F. Crosstalk and Misalignment

• At −20 dB crosstalk, waveguide multiplexing falls to 6.3 bit/s/Hz for co-located users and 6.6 bit/s/Hz for separated users; beyond about −19 dB, switching is better for both.A 10° phase error leaves 6.1 bit/s/Hz for co-located users and 8.0 bit/s/Hz for separated users; multiplexing stops at a ceiling as power grows when either error is present.Source: F. Crosstalk and Misalignment · p. 16, Figure 8 · p. 17

G. Rate Retention

• On the uplink, inter-antenna radiation costs schemes that keep both pinches active 3–4% to first order; one-active-pinch TDMA loses nothing.The benchmark uses 20 dBm per user and identical pinches with δ2 = 1/2.Source: G. Rate Retention · pp. 17–18
• At 0.2 mm position error, NOMA retains 58% and waveguide multiplexing 61–70%, while one-active-pinch TDMA and waveguide division retain rates; 1 ms guard time costs both per-slot-reconfiguring schemes 20%.At 100 MHz, per-slot re-pinching and one-active-pinch TDMA keep their rates, while schemes with both pinches active lose 4–7% with OFDM and 70% with the one-tap receiver.Source: G. Rate Retention · p. 17

V. CONCLUSION

• The two-user benchmark finds no universal ranking: the best scheme depends on operating point, hardware, and service target, and must be chosen with placement, channel estimation, and receiver.The comparison remains to be extended to more users and channel-estimation cost; results also await measurement validation.Source: §V · p. 18