Space-Time Adaptive Processing in Connected and Automated Vehicular Radar Platoons

Zahra Esmaeilbeig, Kumar Vijay Mishra, Mojtaba Soltanalian

Abstract

In this study, we develop a holistic framework for space-time adaptive processing (STAP) in connected and automated vehicle (CAV) radar systems. We investigate a CAV system consisting of multiple vehicles that transmit frequency-modulated continuous-waveforms (FMCW), thereby functioning as a multistatic radar. Direct application of STAP in a network of radar systems such as in a CAV may lead to excess interference. We exploit time division multiplexing (TDM) to perform transmitter scheduling over FMCW pulses to achieve high detection performance. The TDM design problem is formulated as a quadratic assignment problem which is tackled by power method-like iterations and applying the Hungarian algorithm for linear assignment in each iteration. Numerical experiments confirm that the optimized TDM is successful in enhancing the target detection performance.

Introduction

In a vehicular platoon, multiple cars travel in close proximity, each equipped with its own radar sensors. While a single vehicle's radar might struggle with obstructions, signal fading, or poor radial velocity alignment with a target, a network of vehicles—acting as a "multistatic" radar—can theoretically provide a much more comprehensive view of the environment.

Multistatic radar scenario in a CAV platoon showing potential interference and target detection challenges.

However, a significant challenge arises: if all vehicles transmit simultaneously using similar Frequency-Modulated Continuous-Wave (FMCW) signals, they create massive mutual interference. Standard Space-Time Adaptive Processing (STAP)—a technique used to filter out clutter and interference—can degrade significantly if transmitters are not carefully coordinated. This research addresses the challenge of transmitter scheduling: determining the optimal firing sequence for antennas across the platoon to maximize collective detection capabilities.

Methodology

The proposed framework utilizes a decentralized multistatic target detection approach. Key technical components include:

  1. System Modeling: A platoon of $K$ vehicles, each with $N$ transmit and $M$ receive antennas, is modeled. Vehicle-to-Vehicle (V2V) communication allows the sharing of positioning and velocity data, enabling the lead vehicle to process signals from the entire platoon as a single large distributed array.
  2. TDM Scheduling: To mitigate mutual interference, Time Division Multiplexing (TDM) is employed, ensuring only one antenna in the entire platoon transmits during any specific pulse.
  3. Optimization via QAP: The scheduling goal is to maximize the mean of the detection test statistic. This is formulated as a Quadratic Assignment Problem (QAP), which is NP-hard.
  4. Iterative Algorithm: To solve the QAP efficiently, the researchers adapt a "power method-like" iterative algorithm. This technique decomposes the complex QAP into a series of simpler Linear Assignment Problems (LAP), which are then solved using the Hungarian (Munkres) algorithm with $O(L^3)$ complexity.

Results and Discussion

The framework was evaluated through numerical simulations modeled after 77 GHz automotive radar specifications. The primary metric for success was the Receiver Operating Characteristic (ROC) curve, plotting the probability of detection ($P_D$) against the probability of false alarm ($P_{FA}$).

ROC curves demonstrating the performance gains of optimized TDM scheduling over sequential transmission as the number of vehicles (K) increases.

As shown in the results, optimized TDM scheduling significantly outperforms standard sequential transmission. Furthermore, the benefits of cooperation scale with the number of vehicles; as the platoon grows from $K=1$ to $K=3$ vehicles, detection performance increases dramatically when optimized scheduling is applied. The framework demonstrates robustness against high sidelobes and interference that typically degrade standard radar performance.

Conclusion and Future Work

This study demonstrates that treating a CAV platoon as a distributed multistatic radar through optimized scheduling significantly enhances target detection. While the results are promising, future work must address real-world constraints such as perfect V2V synchronization and communication latency. Additionally, testing the framework in dense urban environments with complex clutter and exploring hybrid waveform diversity techniques remain important areas for further investigation.

Summary updated

Abstract

In this study, we develop a holistic framework for space-time adaptive processing (STAP) in connected and automated vehicle (CAV) radar systems. We investigate a CAV system consisting of multiple vehicles that transmit frequency-modulated continuous-waveforms (FMCW), thereby functioning as a multistatic radar. Direct application of STAP in a network of radar systems such as in a CAV may lead to excess interference. We exploit time division multiplexing (TDM) to perform transmitter scheduling over FMCW pulses to achieve high detection performance. The TDM design problem is formulated as a quadratic assignment problem which is tackled by power method-like iterations and applying the Hungarian algorithm for linear assignment in each iteration. Numerical experiments confirm that the optimized TDM is successful in enhancing the target detection performance.
BibTeX
@inproceedings{esmaeilbeig2024space,
  author = {Zahra Esmaeilbeig and Kumar Vijay Mishra and Mojtaba Soltanalian},
  title = {{Space-Time Adaptive Processing in Connected and Automated Vehicular Radar Platoons}},
  booktitle = {IEEE International Conference on Acoustics, Speech and Signal Processing},
  year = {2024},
  url = {https://openreview.net/forum?id=LyN26D1Jhz},
}
Figure