Performance Evaluation of Non-GPS Based Localization Techniques under Shadowing Effects

Ngoc Mai Nguyen, Le Chung Tran, Farzad Safaei, Son Lam Phung, Peter Vial, Nam Huynh, Anne Cox, Theresa Harada, Johan BarthélemySensors 2019 · DOI

TL;DR

This paper investigates the performance of non-GPS localization techniques, specifically combinations of Received Signal Strength Indicator (RSSI) and Angle-of-Arrival (AOA), under shadowing effects common in urban environments. The authors propose a mathematical model for shadowing and demonstrate that while AOA is more robust than RSSI, combining them with a higher weight on AOA can significantly improve localization accuracy and precision, especially when adaptive power control is employed.

What Problem Does This Paper Address?

Global Positioning System (GPS) is the go-to for location services, but it falters in environments with obstructed signals, such as urban canyons, indoors, or underwater. This limitation is particularly problematic for Vehicular Ad Hoc Networks (VANETs), where precise vehicle positioning is crucial for safety (collision avoidance) and efficiency (congestion prediction). While non-GPS localization methods like RSSI and AOA exist, their performance under real-world shadowing conditions, where signals are intermittently blocked by obstacles, is not fully understood, especially when these methods are combined. This paper aims to fill that gap by analyzing the impact of shadowing on weighted combinations of RSSI and AOA.

Methods

The paper focuses on range-based non-GPS localization techniques, specifically RSSI and AOA.

  • RSSI: This method estimates distance based on the received signal strength, which is inversely proportional to distance in ideal conditions. However, it's highly susceptible to shadowing. The paper uses a path loss model that incorporates a Gaussian random variable to represent shadowing effects.
  • AOA: This method estimates the angle of arrival of a signal using antenna arrays. The Multiple Signal Classification (MUSIC) algorithm is employed for AOA estimation, which is generally more resilient to noise and shadowing than RSSI.
  • Weighted Combinations: The core of the paper is the proposal and evaluation of various weighted combinations of RSSI and AOA, represented as $p \cdot \text{AOA} + q \cdot \text{RSSI}$, where $p$ and $q$ are weighting factors. Different combinations are explored, varying the number of anchors and the balance between AOA and RSSI contributions.
  • Shadowing Model: A mathematical model is developed to simulate shadowing effects, treating them as Gaussian random variables with zero mean and a defined variance ($\delta^2$). This model is integrated into the localization algorithms for evaluation.

Datasets

Not discussed in the paper. The study relies on simulations rather than real-world datasets.

Key Insights and Ideas

  • Shadowing is a significant challenge: The paper clearly demonstrates that shadowing effects substantially degrade the accuracy and precision of localization techniques, particularly those relying heavily on RSSI.
  • AOA is more robust than RSSI: Angle-of-Arrival (AOA) measurements are shown to be less susceptible to shadowing compared to RSSI measurements. This resilience is a key finding that drives the effectiveness of combined methods.
  • Combining methods is beneficial, but weighting matters: While combining RSSI and AOA generally outperforms using either method alone, the way they are combined is critical. Increasing the weight of the AOA component and decreasing that of RSSI leads to improved performance, especially at higher Signal-to-Noise Ratios (SNRs).
  • More anchors don't always mean better: Counterintuitively, the study found that increasing the number of anchors doesn't automatically guarantee improved precision or accuracy. This suggests that the quality of the measurements and the algorithm's ability to handle them are more important than sheer quantity.
  • Adaptive power control is a promising solution: The paper suggests that adaptive power control algorithms, which automatically increase transmission power in shadowing environments, could significantly enhance the performance of AOA-dominant localization methods, making them suitable for challenging urban scenarios.

Key Results

  • Precision: In ideal conditions (no shadowing), the 2AOA method achieved the best precision. However, under shadowing, methods with higher AOA weighting (like 2AOA and 2AOA + 1RSSI) showed significantly better precision compared to pure RSSI methods.
  • Accuracy: The 2AOA and 2AOA + 1RSSI methods consistently demonstrated the highest accuracy, especially at low SNRs and in the presence of shadowing. For instance, at an RDE of 20%, 2AOA achieved 93% accuracy without shadowing and 83% with shadowing ($\delta^2=1$).
  • Shadowing Impact: Increasing shadowing variance ($\delta^2$) from 0 to 1 and then to 4 led to a noticeable deterioration in localization error for most methods. Distance estimations were found to be more susceptible to shadowing than angle estimations.
  • SNR Influence: While RSSI-based methods showed saturation at higher SNRs under shadowing, AOA-based methods generally improved with increasing SNR.

Limitations and Future Work

  • Limitations:
    • The study relies on simulations, and real-world validation would be beneficial.
  • The paper acknowledges that the MUSIC algorithm's performance can degrade with correlated noise, which is a realistic scenario not fully captured by the ideal noise model.
  • The performance of weighted combinations at lower SNRs showed a trade-off between accuracy and precision, with 2AOA and 2AOA + 1RSSI having higher accuracy but lower precision compared to methods like 1AOA + 1RSSI.
  • Future Work:
    • Investigating the effect of multipath propagation, which also significantly degrades localization performance in VANETs.
  • Analyzing weighted combination algorithms in correlated fading channels between antenna elements.
  • Exploring the use of multi-antenna Orthogonal Frequency Division Multiplexing (OFDM) techniques for non-GPS localization in multipath environments.
  • The authors strongly recommend adaptive power control algorithms to boost the performance of AOA-dominant methods in shadowing environments.

Summary updated

BibTeX
@article{nguyen2019performance,
  author = {Ngoc Mai Nguyen and Le Chung Tran and Farzad Safaei and Son Lam Phung and Peter Vial and Nam Huynh and Anne Cox and Theresa Harada and Johan Barthélemy},
  title = {{Performance Evaluation of Non-GPS Based Localization Techniques under Shadowing Effects}},
  journal = {Sensors},
  year = {2019},
  doi = {10.3390/s19112633},
}
Figure