PaperlayerView exact revision on arXiv

Source-linked AI summary

Integrated Sensing and Communication-assisted Orthogonal Time Frequency Space Transmission for Vehicular Networks

Weijie Yuan, Zhiqiang Wei, Shuangyang Li, Jinhong Yuan, Derrick Wing Kwan Ng

arXiv:2105.03125v2eess.SP

TL;DR

High-mobility vehicular channels challenge reliable OTFS transmission and create overhead for beam alignment and uplink channel estimation. The paper uses OTFS-ISAC echoes to estimate and predict vehicle and channel states, enabling downlink pre-equalization and guard-space-free uplink processing; simulations demonstrate benefits across the proposed scheme.

  • Problem

    High-mobility vehicular networks face time-varying channels, while conventional beam tracking and uplink channel estimation require signaling, pilots, or guard space.

  • Method

    The RSU uses OTFS-ISAC signals and reflected echoes to estimate vehicle states, predict channel parameters, pre-equalize downlink transmission, place uplink symbols without guard space, and detect data with uncertainty-aware SPA processing.

  • Results

    Numerical simulations demonstrate the benefits of the proposed ISAC-assisted OTFS transmission scheme, with proposed uplink estimation and detection approaching ideal perfect-CSI performance.

  • Takeaways & Limitations

    The scheme supports joint sensing and communication while reducing downlink channel-estimation and beam-pairing requirements and lowering uplink training overhead.

Abstract

from arXiv · show

Orthogonal time frequency space (OTFS) modulation is a promising candidate for supporting reliable information transmission in high-mobility vehicular networks. In this paper, we consider the employment of the integrated (radar) sensing and communication (ISAC) technique for assisting OTFS transmission in both uplink and downlink vehicular communication systems. Benefiting from the OTFS-ISAC signals, the roadside unit (RSU) is capable of simultaneously transmitting downlink information to the vehicles and estimating the sensing parameters of vehicles, e.g., locations and speeds, based on the reflected echoes. Then, relying on the estimated kinematic parameters of vehicles, the RSU can construct the topology of the vehicular network that enables the prediction of the vehicle states in the following time instant. Consequently, the RSU can effectively formulate the transmit downlink beamformers according to the predicted parameters to counteract the channel adversity such that the vehicles can directly detect the information without the need of performing channel estimation. As for the uplink transmission, the RSU can infer the delays and Dopplers associated with different channel paths based on the aforementioned dynamic topology of the vehicular network. Thus, inserting guard space as in conventional methods are not needed for uplink channel estimation which removes the required training overhead. Finally, an efficient uplink detector is proposed by taking into account the channel estimation uncertainty. Through numerical simulations, we demonstrate the benefits of the proposed ISAC-assisted OTFS transmission scheme.

I. INTRODUCTION

The paper motivates integrating sensing and communication with OTFS for high-mobility vehicular networks, then proposes an ISAC-assisted design covering both downlink and uplink transmission.

  • Motivation: ISAC integrates sensing and communication by sharing signal processing, hardware architectures, and frequency spectrum.The integrated approach is presented as a way to reduce hardware and spectrum costs.
  • Motivation: High vehicle mobility breaks OFDM subcarrier orthogonality, motivating modulation resilient to time-varying channels.Vehicular environments may involve speeds of 120–200 km/h.
  • OTFS and ISAC: OTFS places data symbols in the delay-Doppler domain and offers resilience against time-varying channels.The symbols are spread across the time-frequency domain after two-dimensional processing.
  • Proposed scheme: The proposed scheme uses OTFS signals simultaneously for vehicle-state sensing and downlink communication, reducing hardware cost and spectral-resource use.Vehicle states are estimated from reflected echoes at the RSU.
  • Downlink transmission: Predicted downlink beamforming compensates channel effects at the RSU, enabling direct single-tap maximum-likelihood detection without vehicle-side channel estimation.Beamformers use estimated kinematic parameters and predicted vehicle states.
  • Uplink transmission: For uplink transmission, predicted delay-Doppler paths enable guard-space-free symbol placement and factor-graph or SPA detection that accounts for channel-estimation uncertainty.The design targets lower training overhead while retaining channel-estimation and detection functionality.

B. Radar Signal Model

The RSU transmits a multibeam ISAC signal to multiple vehicles and processes their reflected echoes to estimate vehicle-specific sensing parameters. Massive-MIMO spatial separation and matched filtering support distinguishing vehicles and estimating their delays and Dopplers.

  • Signal transmission: The RSU prepares a P-dimensional multibeam ISAC signal, with each component carrying information for one vehicle.The signal is transmitted across all Nt antennas through a beamforming matrix F.
  • Signal transmission: Each beamforming-matrix column steers its vehicle-specific signal toward the intended vehicle direction using an allocated power and steering vector.Because the intended direction is unknown, the beam is typically formed using the vehicle’s predicted angle relative to the RSU.
  • Echo reception: The transmitted ISAC signal reflects from moving vehicles, producing echoes characterized by reflection coefficients, round-trip delays, and round-trip Doppler spreads.The RSU receives these echoes through a radar sensing channel that is time- and frequency-selective, with additive white Gaussian noise.
  • Echo separation: Massive-MIMO receive steering vectors for different vehicles are asymptotically orthogonal, allowing the RSU to distinguish their individual echoes.The separated echoes are represented as vehicle-specific components before matched filtering.
  • Parameter estimation: Matched filtering uses a bank of transmitted vehicle signals over candidate delay and Doppler values to estimate each vehicle’s signaling delay and Doppler shift.A peak occurs when the filter parameters match the corresponding vehicle echo; the matched filter also provides an SNR gain related to the signal energy.

D. Communication Model

The communication model describes OTFS downlink and uplink input-output relationships in the delay-Doppler domain. Downlink transmission uses predicted channel parameters for beamforming, while uplink reception accounts for multipath delays, Dopplers, angles, and DD-domain convolution.

  • Downlink model: The downlink channel is modeled as line-of-sight dominated under asymptotically orthogonal steering vectors for different directions.The received downlink signal includes additive noise in the delay-Doppler domain.
  • OTFS processing: The DD-domain processing chain obtains TF samples through receive filtering and then transforms them into DD samples using the symplectic finite Fourier transform.The signal model includes additional Gaussian noise with PSD N0.
  • Downlink model: Radar and communication parameters differ because radar delays and Dopplers describe round-trip echoes, whereas downlink parameters describe communication signals; the round-trip Doppler is generally 2νi.The symbols γi and ωi denote radar delay and Doppler, while τi and νi denote downlink communication quantities.
  • Downlink model: The downlink DD-domain input-output relationship shifts transmitted symbols by the vehicle’s Doppler and delay indices and applies the beamforming channel gain.The model assumes ideal transmit and receive filtering and sufficient Doppler resolution with integer Doppler indices.
  • Uplink model: The uplink vehicle-to-RSU channel is multipath because a single-antenna vehicle’s signal can scatter from other vehicles in the network.Each path is described by a channel gain, delay, Doppler, and angle; the direct path has zero delay.
  • Uplink model: After OTFS modulation, receive beamforming, multicarrier demodulation, filtering, and SFFT, the uplink signal is represented in the DD domain.The path delays and Dopplers map to delay and Doppler indices, and the DD input-output relationship can be expressed as a two-dimensional convolution with an effective channel.

III. ISAC-ASSISTED OTFS COMMUNICATIONS

The paper’s proposed ISAC-assisted OTFS scheme combines sensing, vehicular-topology prediction, and communication receiver design for uplink and downlink transmissions.

  • III. ISAC-ASSISTED OTFS COMMUNICATIONS: The proposed framework first studies sensing-parameter estimation and communication-channel prediction before designing uplink and downlink receivers.The framework is organized around sensing assistance for both communication directions.

A. General Framework for ISAC-assisted OTFS Communications

The framework uses OTFS-ISAC echoes to estimate vehicle states, predict the next network topology, and assist both downlink beamforming and uplink channel symbol placement. This connects sensing-based state prediction directly to communication design.

  • State estimation: At time instant η, the RSU transmits OTFS-ISAC signals and estimates vehicles’ delays, Dopplers, and angles from their reflected echoes.These quantities provide the sensing inputs for subsequent state processing.
  • Dynamic topology construction and prediction: The RSU infers vehicle locations and speeds from the estimated sensing parameters, constructs the dynamic vehicular topology, and predicts states at η + 1.The prediction concerns the vehicles’ speeds and locations at the following time instant.
  • Assistance to downlink communication: Using predicted vehicle states, the RSU predicts future angles and channel impairments and forms downlink beamformers before transmitting at η + 1.The beamformers are designed from predicted parameters to combat the anticipated channel impairments.
  • Assistance to uplink communication: For uplink assistance, predicted locations and speeds yield relative vehicle distances and speeds, which are converted into uplink DD-domain delays and Dopplers.Knowing the multipath interference pattern enables a new symbol-placement scheme with much lower training overhead.

B. Sensing Parameter Estimation and Prediction

The RSU estimates vehicle sensing parameters from reflected echoes, converts them into locations and speeds, and predicts vehicle motion and angles for subsequent transmission.

  • Sensing parameter estimation: The RSU estimates vehicle angles, delays, Dopplers, locations, and speeds from reflected echoes.These estimates support subsequent prediction of vehicle states and communication parameters.
  • State prediction: Vehicle locations and speeds are predicted for the next OTFS frame using the estimated motion parameters.The prediction uses the OTFS frame duration and assumes vehicle speeds remain unchanged over the short interval.
  • Beam alignment prediction: The RSU predicts each vehicle’s angle relative to the RSU and uses it to formulate the next transmit beamformer.The prediction-based protocol avoids the conventional sequence of dedicated pilots, vehicle feedback, and subsequent beamforming.

C. Downlink Communication

For downlink transmission, the RSU uses predicted channel parameters to pre-equalize the signal, allowing vehicles to detect data without channel estimation.

  • Pre-equalization: Pre-equalization lets vehicles bypass downlink channel estimation after receiving the OTFS-ISAC signal.The approach compensates the downlink channel effect at the RSU side.
  • Pre-equalization: The RSU obtains the predicted channel gain and Doppler for each vehicle from predicted motion parameters.These quantities are used to compensate channel attenuation and Doppler at the transmitter.
  • Data detection: A single-tap maximum-likelihood detector can infer transmitted symbols from the approximately equalized received samples.The detector selects symbols from the transmitted constellation set A.
  • Beam alignment: Maximum receive SNR is achieved when the predicted beam angle matches the actual vehicle angle.Angular prediction accuracy directly affects the antenna array gain available to the vehicle.

D. Uplink Communication

For uplink transmission, the RSU predicts the delay and Doppler of multipath components from the dynamic vehicular topology and uses these predictions to characterize interference.

  • Multipath prediction: The uplink contains multiple propagation paths when vehicles use single antennas.The RSU predicts each path’s delay and Doppler using vehicle locations, speeds, and angles.
  • Multipath prediction: Predicted locations, speeds, and angles allow the RSU to determine delays and Dopplers for reflected uplink paths.For a path reflected by another vehicle, the parameters follow the vehicles’ geometric relationships.
  • Channel and interference characterization: The direct-path uplink Doppler is the reverse of the downlink Doppler, while sufficient delay-Doppler resolution makes the interference pattern known to the RSU.Receive beamforming for a reflected path can use the predicted angle of the reflecting vehicle.

1) Uplink Channel Estimation:

The proposed uplink channel-estimation scheme predicts delay-Doppler support and removes guard-space requirements, eliminating the associated training overhead while modeling estimation uncertainty.

  • Conventional channel estimation: Conventional OTFS estimation places one pilot in an OTFS frame and reserves guard space sized by the maximum delay and Doppler indices.The guard space prevents pilot spreading from interfering with data symbols.
  • Proposed symbol placement: The proposed scheme uses predicted delay and Doppler indices to place symbols without guard space.This placement permits pilot-data interference, which is handled during channel estimation.
  • Estimation uncertainty: The proposed channel estimate remains uncertain because received samples contain contributions from data symbols, interference, and noise.The uncertainty is quantified through the uplink input-output relationship.
  • Estimation uncertainty: A higher pilot-to-data power ratio produces smaller channel-estimation uncertainty.The pilot power is denoted by Ep.

2) Data Detection for Uplink Transmission:

The uplink detector represents the received-sample distribution with a factor graph and applies SPA message passing while incorporating channel-estimation uncertainty. This produces probability-based symbol decisions or decoder inputs for coded systems.

  • Symbol decisions: For coded systems, constellation probabilities generate bit LLRs for channel decoding; uncoded systems select symbols by comparing those probabilities.The prior distribution is obtained from decoder LLRs for coded systems and from constellation probabilities for uncoded systems.
  • Factor-graph detection: The joint distribution of uplink samples and transmitted symbols is represented by a factor graph for symbol detection.The graph connects variable nodes representing symbols with function nodes representing received samples.
  • Factor-graph detection: SPA updates messages between variable and function nodes to approximate each transmitted symbol's marginal distribution.The detector uses variable-to-function and function-to-variable messages, then combines them to obtain symbol probabilities.
  • Uncertainty-aware detection: The detector revises message variances using channel-estimation uncertainty rather than treating the estimated channel as exact.The revised variance accounts for the uncertainty in the estimated channel tap and transmitted symbol product.
  • Uncertainty-aware detection: Including channel-uncertainty information in the message derivation improves detection performance.

IV. SIMULATION RESULTS

Simulations evaluate sensing, beam alignment, channel estimation, training overhead, and BER in a four-vehicle high-mobility scenario. The proposed prediction-based OTFS-ISAC scheme improves tracking and communication while eliminating uplink training overhead.

  • Simulation setup: The simulation uses four vehicles on a two-lane road, with speeds uniformly drawn from 10–15 m/s, a 3 GHz carrier, and 6 kHz subcarrier spacing.
  • Sensing performance: The angle-estimation RMSE sharply decreases over the first time instants and reaches 10^-2 rad across the considered antenna-array configurations.The configurations use Nt = Nr = 16, 32, 64, and 128.
  • Downlink communication: The proposed prediction-based beam alignment significantly outperforms feedback-based and auxiliary beam-pairing methods in receive SNR.The receive SNR initially rises as the vehicle approaches the RSU and then falls as it moves away.
  • Downlink communication: The proposed downlink algorithm approaches perfect-CSI BER performance, while conventional beam pairing and channel estimation lose BER performance because of degraded receive SNR.The proposed approach also requires no pilots for downlink communication.
  • Uplink channel estimation: 12.5% of 3840 delay-Doppler grids are reserved for conventional channel estimation, whereas the proposed symbol placement uses all grids for data and has zero training overhead.The proposed estimate's NMSE is bounded by approximately 10^-2 under the stated pilot-power condition.
  • Uplink communication: The proposed uplink channel-estimation and detection scheme approaches perfect-CSI and conventional-OTFS BER performance and improves BER over uncertainty-neglecting SPA detection.OFDM MMSE detection degrades severely because subcarrier orthogonality does not hold in vehicular scenarios.

V. CONCLUSIONS

The paper proposes ISAC-assisted OTFS transmission for vehicular networks, using sensed vehicle motion to predict states and channel parameters. Simulations show reduced training overhead with reliable downlink and uplink communications.

  • Conclusion: The RSU estimates vehicle motion from reflected echoes and predicts vehicle states and delay-Doppler channel parameters.
  • Conclusion: Downlink transmission requires neither dedicated channel-estimation pilots nor beam-pairing pilots.
  • Conclusion: The guard-space-free uplink symbol-placement scheme yields much lower channel-estimation training overhead than the conventional approach.
  • Conclusion: Simulation results show reduced training overhead while maintaining reliable communications.
Loading 2105.03125v2…