PaperlayerView exact revision on arXiv

Source-linked AI summary

A Continuous Payload-Bearing Discrete Multitone Modulation Framework for Fiber-Optic Integrated Sensing and Communication

Huan Huang, Ziang Chen, Zhiyang Xue, Dongdong Zou, Yi Cai

arXiv:2608.29020v1eess.SP

TL;DR

Fiber-optic ISAC needs a payload-bearing waveform that simultaneously supports communication and distributed sensing without sensing-only silence. The paper develops continuous CP-DMT and NoCP-DMT reconstruction modes, demonstrating 600-Hz localization near 5.071 km with low EVM and no bit errors.

  • Problem

    Existing approaches do not use payload symbols themselves as the known excitation for backward distributed-channel reconstruction, while pulse-based sensing requires RTT-scale silence and suffers spatial ISI.

  • Method

    The framework uses one continuous payload-bearing DMT waveform for forward IM/DD communication and backward DAS, with CP-DMT using full-memory CP and FDE and NoCP-DMT using regularized LS.

  • Results

    The experiments localized a 600-Hz disturbance at 5.071 km, achieved gauge-phase correlations of 0.989 and 0.987, and obtained 4.80%–5.00% EVM with no bit errors.

  • Takeaways & Limitations

    Both waveform modes support forward payload recovery and backward DAS, while NoCP-DMT trades cyclic redundancy for higher long-memory reconstruction complexity.

Abstract

from arXiv · show

A key challenge in fiber-optic integrated sensing and communication (ISAC) is to make the payload-bearing waveform itself serve both functions without a separate sensing waveform or sensing-only silent interval. We propose a continuous discrete multitone (DMT) framework with two waveform modes, in which the same payload-bearing waveform supports forward intensity-modulation/direct-detection (IM/DD) communication and backward distributed acoustic sensing (DAS). A unified phase-sensitive optical time-domain reflectometry (φ-OTDR) model represents distributed Rayleigh backscattering as a finite-memory sensing multipath channel. It shows that conventional pulse-and-wait φ-OTDR requires a round-trip-time-scale silent interval to isolate successive returns, while nonzero off-peak samples in practical matched-filter (MF) pulse compression cause spatial intersymbol interference (ISI). Continuous DMT instead retains superposed returns and separates range-cell contributions through known-waveform channel reconstruction. Cyclic-prefix DMT (CP-DMT) uses a full-memory CP and one-tap frequency-domain equalization (FDE); under sufficient-CP and ideal full-bin inversion conditions, it eliminates MF-correlation-induced spatial ISI. Cyclic-prefix-free DMT (NoCP-DMT) applies regularized least squares (LS) to the long-memory linear convolution, avoiding the CP at higher receiver complexity. Experiments over a 10-km fiber link localize a 600-Hz disturbance applied by a piezoelectric transducer (PZT) near 5.071 km and recover gauge-differential phase with correlations of 0.989 and 0.987 for CP-DMT and NoCP-DMT, respectively. At 2- and 1-V PZT drive levels, the ten-record localization standard deviations are 0.16/0.11 m and 0.18/0.38 m, respectively. The corresponding IM/DD error vector magnitude values range from 4.80% to 5.00%, with no bit errors observed.

I. INTRODUCTION

Fiber-optic ISAC seeks to use a common payload-bearing waveform for communication and distributed sensing. This work develops continuous DMT modes that support forward IM/DD communication and backward DAS while addressing return overlap and spatial ISI.

  • Fiber-optic ISAC combines high-capacity data transport with sensitivity to distributed strain, vibration, temperature, and acoustic disturbances.
  • Existing fiber-optic ISAC approaches separate functions through wavelength, frequency, space, receiver processing, or structured waveform components rather than payload symbols themselves.
  • Conventional matched-filter sensing requires an RTT-scale silent interval and still suffers spatial ISI from nonzero off-peak correlation samples.
  • The proposed framework uses one continuous payload-bearing DMT waveform for forward IM/DD payloads and backward DAS excitation.
  • CP-DMT uses full-memory-CP-assisted FDE, whereas NoCP-DMT uses regularized long-memory reconstruction and removes cyclic redundancy at higher complexity.
  • Experiments localized a 600-Hz disturbance near 5.071 km, recovered gauge-phase correlations of 0.989 and 0.987, and achieved 4.80%–5.00% EVM with no bit errors.

II. PHASE-SENSITIVE OPTICAL TIME-DOMAIN REFLECTOMETRY-BASED DISTRIBUTED ACOUSTIC SENSING

This section establishes a general φ-OTDR model and identifies the guard requirement and spatial ISI associated with matched-filter LFM pulse compression.

  • The section develops a general φ-OTDR model from coherent optical reception to a discrete finite-memory distributed sensing channel.
  • It then specializes the model to matched-filter pulse compression, identifying RTT-driven guard requirements and spatial ISI.

A. Phase-Sensitive Optical Time-Domain Reflectometry

The φ-OTDR formulation maps coherent Rayleigh backscatter onto a time-varying finite-memory sensing channel whose taps correspond to physical range cells. Within quasi-static observation intervals, known excitation and received samples form a linear convolution model.

  • Heterodyne reception and digital downconversion produce a complex observation containing the amplitude and phase information needed for distributed-channel modeling.
  • The distributed Rayleigh response is modeled as a time-varying channel whose delay maps to physical range through z = v_gτ/2.
  • When the channel remains approximately constant over an observation interval, pulsed and continuous probing follow the same finite-memory input–output model.
  • The channel snapshot includes Rayleigh scattering, attenuation, static phase, and perturbation-induced phase, whose slow-time evolution carries acoustic information.
  • The discrete model uses a minimum RTT-covering support order M_min = ⌈T_RTT/T_s⌉ + 1, with each nonzero tap representing coherent backscatter from a physical range cell.

B. Phase-Sensitive Optical Time-Domain Reflectometry Using MF-Based LFM Pulse Compression

Matched-filter LFM pulse compression trades waveform duration for energy and spatial resolution but requires RTT-scale return isolation and leaves correlation-induced spatial ISI. These effects constrain payload duty cycle, localization fidelity, and slow-time bandwidth.

  • MF-based LFM pulse compression transmits a long broadband pulse and applies matched filtering to improve the energy–spatial-resolution tradeoff.
  • Adjacent pulse returns are isolated when the silent guard satisfies T_g ≥ T_RTT, equivalently T_PRI ≥ T_LFM + T_RTT.
  • The matched-filter output convolves the distributed channel with a waveform-dependent correlation kernel rather than directly inverting the sensing channel.
  • Nonzero off-peak correlation samples couple different range-bin taps, producing spatial ISI that can leak strong responses into neighboring bins and distort phase traces.
  • Extending the silent guard prevents inter-pulse interference but cannot remove spatial ISI, which is governed by the pulse-compression kernel.
  • RTT-scale silent intervals reduce payload duty cycle, while pulse-repetition sampling limits the unaliased acoustic bandwidth to 1/(2T_PRI).

III. CONTINUOUS DISCRETE MULTITONE MODULATION FRAMEWORK FOR ISAC

The framework uses one continuous payload-bearing DMT waveform for both forward IM/DD communication and backward DAS. It develops CP-DMT and NoCP-DMT waveform modes for continuous waveform-level ISAC.

  • One continuous payload-bearing DMT waveform supports forward communication and backward DAS.
  • The framework includes CP-DMT and NoCP-DMT modes for continuous waveform-level ISAC.

A. Phase-Sensitive Optical Time-Domain Reflectometry Using Frequency-Domain Channel Reconstruction

CP-DMT reconstructs the distributed sensing channel in the frequency domain using a full-memory cyclic prefix, while continuous transmission avoids a sensing-only guard interval. Under ideal conditions, reconstruction removes MF-induced spatial ISI, but practical bandwidth limits, CP overhead, and slow-time constraints remain.

  • System and signal model: CP-DMT uses frequency-domain channel reconstruction with a known transmit-derived reference and pilot-aided one-tap FDE for the forward IM/DD link.The sensing receiver uses heterodyne detection and continuous payload-bearing blocks without a sensing-only interblock guard.
  • CP-DMT channel model: A sufficient full-memory CP makes the retained linear convolution equivalent to N-point circular convolution when M ≤N.The CP supplies current-block samples required by channel memory and confines preceding-block interference to discarded CP samples.
  • Ideal and practical reconstruction: Under sufficient CP, exact alignment, quasi-static sensing, and noiseless full-bin unregularized inversion, CP-DMT exactly recovers grid-aligned taps without MF-correlation-induced spatial ISI.The practical numerical response instead includes spectral selection, missing-bin interpolation, and regularization.
  • Ideal and practical reconstruction: Inactive or unreliable subcarriers, finite reconstruction bandwidth, and regularization broaden the periodic reconstruction kernel; approximately, Δzres ≈vg/(2Brec).The reconstructed taps feed subsequent gauge-differential phase recovery.
  • Overhead and slow-time constraints: Continuous CP-DMT replaces the RTT-scale silent guard with cyclic redundancy but retains full-memory CP overhead and slow-time limits.Usable acoustic bandwidth is jointly limited by slow-time aliasing, the block-aperture response, and the quasi-static-channel condition.
  • CP-DMT waveform structure: Each CP-DMT block repeats Ncp samples around an N-sample useful DMT symbol, with no zero-valued interblock guard interval.

B. Cyclic-Prefix-Free Continuous Discrete Multitone Modulation Waveform

NoCP-DMT removes the repeated full-memory cyclic prefix and reconstructs the long-memory distributed sensing channel directly from known waveform context. This preserves continuous payload transmission while requiring regularized least-squares processing and RTT-scale observation windows.

  • Constraints and trade-offs: The full-memory CP in CP-DMT enables low-complexity frequency-domain equalization, whereas NoCP-DMT uses Ncp = 0 and ηcp = 1.The comparison makes the overhead–complexity trade-off explicit between CP-assisted equalization and cyclic-prefix-free reconstruction.
  • Waveform construction: NoCP-DMT concatenates useful DMT symbols directly, eliminating copied prefixes and zero-valued intersymbol guard intervals while retaining payload-bearing transmission.Removing the cyclic prefix reduces full-memory sample overhead, although forward-link equalization remains separate from backward DAS reconstruction.
  • Waveform construction: Without a cyclic-prefix condition, NoCP-DMT models linear convolution directly rather than replacing it with blockwise circular convolution.Samples transmitted before a DMT-symbol boundary remain part of the distributed sensing-channel model.
  • Long-memory reconstruction: The NoCP observation model includes MLS −1 preceding samples so contributions spanning adjacent DMT symbols are incorporated through complete known input context.The resulting finite-memory model uses a selected LS support order MLS and zero-pads the physical channel when necessary.
  • Long-memory reconstruction: Regularized least squares reconstructs the distributed sensing channel from the finite-memory observation model.The normal-equation solution uses regularization, while FFT-based convolution and correlation can avoid forming or inverting a dense Toeplitz matrix.
  • Constraints and trade-offs: NoCP-DMT removes transmitted cyclic-prefix overhead but not the physical sensing memory or RTT-scale observation requirement.Positive λLS improves stability at the cost of bias, and the observation window must remain within the quasi-static interval assumed by the model.

IV. EXPERIMENTAL RESULTS AND DISCUSSION

The experiments validate continuous CP-DMT and NoCP-DMT over an approximately 10-km fiber link, using the same waveform for forward IM/DD payload transmission and backward DAS. Both modes localize the 600-Hz PZT disturbance near 5.071 km, recover its phase evolution, and maintain payload recovery under the tested conditions.

  • Experimental setup: The setup used the same optical transmitter and approximately 10-km fiber link for continuous CP-DMT and NoCP-DMT experiments.The link included a PZT at the junction of two approximately 5-km fiber spools.
  • Reconstruction and validation: The two modes used distinct reconstruction parameterizations: one-tap sensing-channel FDE for CP-DMT and regularized LS channel reconstruction for NoCP-DMT.CP-DMT used a full-memory CP, whereas NoCP-DMT used a long-memory observation and reconstruction structure without the transmitted CP.
  • Reconstruction and validation: 0.989 and 0.987 were the recovered gauge-differential phase correlations for CP-DMT and NoCP-DMT, respectively, under a 2-V, 600-Hz PZT drive.Both modes localized the disturbance at 5.071 km and recovered a dominant spectral component near 600 Hz.
  • Interpretation and scope: The reported positions are internally consistent rather than externally calibrated absolute-range coordinates because residual hardware skew was not externally removed.The ten-record standard deviations are finite-sample repeatability statistics, and the experiments demonstrate feasibility rather than an all-else-equal performance ranking.
  • Joint sensing and communication: 4.80%–5.00% mean EVM was obtained, with no bit errors in 40 communication records and successful cyclic-redundancy checks for all recovered images.These results show simultaneous forward payload recovery and backward DAS under the tested PZT drives.

V. CONCLUSIONS

The framework demonstrates continuous payload-bearing DMT for joint forward IM/DD communication and backward DAS, using CP-DMT or NoCP-DMT to reconstruct sensing channels. Experiments show localized disturbance sensing, phase recovery, and communication without a separate sensing waveform.

  • CP-DMT uses sufficient CP and FDE, while NoCP-DMT uses regularized LS without a CP but with higher receiver complexity.Under ideal full-bin inversion, CP-DMT avoids MF-induced coupling; finite bandwidth and regularization can still cause practical spreading.
  • 0.989 and 0.987 gauge-differential phase correlations were obtained for CP-DMT and NoCP-DMT near the 5.071-km coordinate.Both modes blindly localized a 600-Hz PZT disturbance over an approximately 10-km link.
  • 0.11–0.38 m localization standard deviations and 4.80%–5.00% mean IM/DD EVM were measured across four ten-record conditions, with no bit errors.These finite samples demonstrate record-to-record repeatability but do not establish drive-voltage dependence.
  • The results establish the feasibility of common-waveform fiber-optic ISAC without a separate sensing waveform.The payload-bearing communication waveform itself serves as the sensing excitation, providing a foundation for access links combining data transmission with distributed monitoring.
Loading 2608.29020v1…