Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Predictive Beamforming with Distributed MIMO

Hasret Taha Akçalı, Özlem Tuğfe Demir, Tolga Girici, Emil Björnson

TL;DR

This work addresses beam tracking overhead in mmWave V2X by extending predictive beamforming to a distributed MIMO ISAC network of RSUs. It derives an EKF-based state evolution to predict vehicle velocity, angle, distance, and radar cross-section across multiple RSUs, enabling coordinated predictive beamforming. Simulations show that distributed MIMO provides more uniform sensing and spectral efficiency than a co-located array with the same total antennas, particularly during vehicle motion, underscoring robustness and coverage benefits. The results highlight the potential of distributed ISAC to enhance sensing accuracy and data-rate stability in V2X mmWave deployments.

Abstract

In vehicle-to-everything (V2X) applications, roadside units (RSUs) can be tasked with both sensing and communication functions to enable sensing-assisted communications. Recent studies have demonstrated that distance, angle, and velocity information obtained through sensing can be leveraged to reduce the overhead associated with communication beam tracking. In this work, we extend this concept to scenarios involving multiple distributed RSUs and distributed MIMO (multiple-input multiple-output) systems. We derive the state evolution model, formulate the extended Kalman-filter equations, and implement predictive beamforming for distributed MIMO. Simulation results indicate that, when compared with a co-located massive MIMO antenna array, distributed antennas lead to more uniform and robust sensing performance, coverage, and data rates, while the vehicular user is in motion.

Predictive Beamforming with Distributed MIMO

TL;DR

This work addresses beam tracking overhead in mmWave V2X by extending predictive beamforming to a distributed MIMO ISAC network of RSUs. It derives an EKF-based state evolution to predict vehicle velocity, angle, distance, and radar cross-section across multiple RSUs, enabling coordinated predictive beamforming. Simulations show that distributed MIMO provides more uniform sensing and spectral efficiency than a co-located array with the same total antennas, particularly during vehicle motion, underscoring robustness and coverage benefits. The results highlight the potential of distributed ISAC to enhance sensing accuracy and data-rate stability in V2X mmWave deployments.

Abstract

In vehicle-to-everything (V2X) applications, roadside units (RSUs) can be tasked with both sensing and communication functions to enable sensing-assisted communications. Recent studies have demonstrated that distance, angle, and velocity information obtained through sensing can be leveraged to reduce the overhead associated with communication beam tracking. In this work, we extend this concept to scenarios involving multiple distributed RSUs and distributed MIMO (multiple-input multiple-output) systems. We derive the state evolution model, formulate the extended Kalman-filter equations, and implement predictive beamforming for distributed MIMO. Simulation results indicate that, when compared with a co-located massive MIMO antenna array, distributed antennas lead to more uniform and robust sensing performance, coverage, and data rates, while the vehicular user is in motion.

Paper Structure

This paper contains 10 sections, 28 equations, 6 figures, 2 tables.

Table of Contents

  1. Introduction
  2. System Model
  3. State Evolution Model
  4. Signal Model
  5. Radar Signal Model
  6. Radar Measurement Model
  7. Communication Model
  8. Proposed Approach
  9. Numerical Results
  10. Conclusions

Figures (6)

  • Figure 1: Illustration of the system model, where the RSUs are connected to a CPU (not shown in the figure).
  • Figure 2: Frame structure of the proposed approach.
  • Figure 3: V2X state evolution model.
  • Figure 4: Actual and predicted parameters by $L=4$ RSUs with $N_t=N_r=16$ antennas, while the vehicle is moving from right to left. Rows correspond to each RSU, while columns correspond to distance and angle.
  • Figure 5: The CDF of the SE for different configurations.
  • ...and 1 more figures