Multiscale Vehicle Localization in Heterogeneous Mobile Communication Networks

Lele Cong, Kaitao Meng, Deshi Li, Hao Jiang, Liang XuIEEE Internet of Things Journal 2025 · OpenReview

TL;DR

This paper introduces a novel "road-aware" localization system for vehicles using heterogeneous mobile communication networks (HetNets). By segmenting roads into adaptive segments and extracting "salient features" from network signals, the system achieves significantly lower latency and higher precision compared to existing methods, crucial for intelligent transportation.

What Problem Does This Paper Address?

Accurate and low-latency vehicle localization is vital for traffic safety and management. While GPS is common, it struggles in urban environments with signal blockage and can have high latency. Other cellular network-based methods like Cell ID, geometric-based, and fingerprinting methods also face challenges: Cell ID offers low precision, geometric methods are degraded by non-line-of-sight (NLoS) conditions, and fingerprinting is computationally intensive and time-consuming. The core problem is the trade-off between localization precision and computational complexity (leading to latency), especially in complex, dynamic urban road networks.

Methods

The proposed solution, a Multi-scale Vehicle Localization (MSVL) algorithm, tackles this by:

  • Road-Aware Localization Mechanism: This framework leverages HetNet signals (like 4G and 5G) to build a multi-scale positioning space.
  • Singular Point (SP) Segmentation: Roads are adaptively partitioned into segments of unequal lengths. This is achieved by identifying "singular points" where signal gradients change significantly, effectively creating a more granular and informative representation of the road.
  • Salient Feature (SF) Extraction: Instead of using all signal data, the method extracts "salient features" – those with significant differences across road segments. This sparse representation reduces computational load and improves matching accuracy.
  • Multi-scale Feature Matching: The MSVL algorithm integrates road and segment features, using inclusion and adjacency relationships to quickly identify the vehicle's road, segment, and finally, its precise coordinates.
  • Coordinate Mapping: For fine-grained positioning within a segment, polynomial curve fitting is used to map signal features to 2D coordinates.
  • Cramér-Rao Lower Bound (CRLB) Analysis: The paper theoretically verifies the improvement in positioning accuracy brought by the proposed segmentation and SF extraction methods.

Datasets

The paper uses both simulated and real-world data for evaluation:

  • Simulated Data: Collected using the Qualnet simulator, representing a 0.6 Km x 0.6 Km area with 1 MBS and 5 SBSs.
  • Real-world Data: Collected from a 0.15 Km x 0.15 Km area at Wuhan University using a Great Scott Gadget HackRF, capturing signals from 2 HetNet BSs.

No new benchmark datasets were introduced; existing methods were compared against the proposed approach.

Key Insights and Ideas

  • Roads as Dynamic Segments: The idea of partitioning roads into adaptive segments based on signal gradient changes is a key innovation. This moves beyond fixed grid-based approaches.
  • Salient Features for Efficiency: Focusing on "salient features" is crucial for reducing computational complexity. It's not just about having more data, but about intelligently selecting the most informative data points.
  • HetNet Synergy: The approach effectively leverages the diverse signals from heterogeneous networks (4G/5G) to overcome individual signal limitations.
  • Theoretical Validation: Using CRLB analysis to mathematically prove the benefits of their segmentation and feature extraction methods adds significant credibility.
  • Multi-scale Representation: The hierarchical approach (road -> segment -> coordinate) allows for efficient localization at different levels of detail.

Key Results

  • Lower Latency: The proposed MSVL algorithm achieved significantly lower mean delay, with results as low as 7.2 ms in real-world tests, well below the 10 ms threshold for high-speed vehicles.
  • Higher Precision: The system demonstrated improved localization accuracy, with RMSE values approaching the theoretical CRLB. For example, in simulated measurements, it showed improvements of up to 77% compared to benchmark methods.
  • Robustness to BS Number: The method performs well even with a limited number of BSs, outperforming benchmarks in scenarios with fewer available signals.
  • Efficiency: The sparse feature representation and multi-scale approach lead to reduced computational complexity, making it suitable for real-time applications.

Limitations and Future Work

  • Faint Signal Variations: The paper acknowledges that faint HetNet signal variations might be a concern for localization performance.
  • Future Work: The authors suggest addressing this by utilizing signals from more HetNet BSs. They also propose exploring high-precision beamforming to achieve accurate lane-level positioning by extracting lane-level signal features.

Summary updated

Abstract

Low-latency and high-precision vehicle localization plays a significant role in enhancing traffic safety and improving traffic management for intelligent transportation. However, in complex road environments, the low latency and high precision requirements could not always be fulfilled due to the high complexity of localization computation. To tackle this issue, we propose a road-aware localization mechanism in heterogeneous networks (HetNet) of the mobile communication system, which enables real-time acquisition of vehicular position information, including the vehicular current road, segment within the road, and coordinates. By employing this multiscale localization approach, the computational complexity can be greatly reduced while ensuring accurate positioning. Specifically, to reduce positioning search complexity and ensure positioning precision, roads are partitioned into low-dimensional segments with unequal lengths by the proposed singular point (SP) segmentation method. To reduce feature-matching complexity, distinctive salient features (SFs) are extracted sparsely representing roads and segments, which can eliminate redundant features while maximizing the feature information gain. The Cramér-Rao lower bound (CRLB) of vehicle positioning errors is derived to verify the positioning accuracy improvement brought from the segment partition and SF extraction. Additionally, through SF matching by integrating the inclusion and adjacency position relationships, a multiscale vehicle localization (MSVL) algorithm is proposed to identify vehicular road signal patterns and determine the real-time segment and coordinates. Simulation results show that the proposed multiscale localization mechanism can achieve lower latency and high precision compared to the benchmark schemes.
BibTeX
@article{cong2025multiscale,
  author = {Lele Cong and Kaitao Meng and Deshi Li and Hao Jiang and Liang Xu},
  title = {{Multiscale Vehicle Localization in Heterogeneous Mobile Communication Networks}},
  journal = {IEEE Internet of Things Journal},
  year = {2025},
  url = {https://openreview.net/forum?id=4FndG8vzcK},
}
Figure