Localization in Modified Polar Representation: Hybrid Measurements and Closed-Form Solution

Xunchao Cong, Yimao Sun, Yanbing Yang, Lei Zhang, Liangyin ChenJournal of Systems Engineering and Electronics 2023 · DOI

This paper presents a more efficient way to pinpoint the location of a radio source by combining two types of data—where the signal is coming from (Angle of Arrival) and when it arrives (Time Difference of Arrival)—using a clever mathematical coordinate system that works regardless of whether the source is near or far.

TL;DR

The authors propose a new closed-form algorithm called Hybrid Constrained Optimization (HCO) that locates radio sources using a "Modified Polar Representation." This approach unifies near-field and far-field localization into a single framework, achieving theoretical peak accuracy (Cramér-Rao Bound) while being significantly faster and more stable than existing iterative or complex optimization methods.

What Problem Does This Paper Address?

In the world of wireless sensor networks (WSNs) and the Internet of Things (IoT), locating a signal source usually relies on Cartesian $(x, y, z)$ or standard Polar coordinates. However, these systems face a "thresholding effect": if a source is very far away, the range becomes nearly impossible to calculate accurately, and the math often fails unless you already know the source is in the "far-field."

Researchers previously introduced Modified Polar Representation (MPR) to solve this. MPR uses angles and inverse-range ($1/r$) as coordinates. Because $1/r$ simply goes to zero as a source moves toward infinity, it handles near and far sources seamlessly. The problem? Current MPR solvers are either computationally "heavy" (like Semidefinite Relaxation), prone to failing/diverging (like Gauss-Newton iteration), or unstable in certain directions (like the HSUM algorithm). The authors sought a "Goldilocks" solution: a closed-form mathematical formula that is fast, accurate, and robust.

Methods

The core of the paper is the HCO (Hybrid Constrained Optimization) algorithm. Here is how it works:

  1. Pseudo-Linearization: The authors transform the non-linear measurement equations (AOA and TDOA) into a "pseudo-linear" form. This makes the complex geometry of intersecting curves look like a simpler system of linear equations.
  2. Two-Stage Optimization:
    • Stage 1 (Coarse): It finds an initial estimate by solving a quadratic optimization problem. It treats the source direction as a unit vector constraint.
  3. Stage 2 (Refinement): Since the first stage ignores some mathematical dependencies to stay simple, the second stage applies a "correction" to the coarse result.
  4. Closed-Form: Unlike iterative methods that "guess and check" until they find a solution, HCO is a direct mathematical path to the answer. This guarantees it won't "diverge" or get stuck in a loop.

Datasets

As is standard in signal processing research, the authors used simulated environments to rigorously control noise levels and geometry:

  • General Scenario: Sensors are placed randomly, and the source is at a fixed range (80m) with varying noise levels.
  • Large Equal Radius (LER) Scenario: A specific, challenging geometry where all sensors are roughly the same distance from the source, making TDOA measurements very small and difficult to process.
  • Benchmarks: They compared their results against several established methods, including HSUM, HGTRS, SDR, and the traditional Cartesian Maximum Likelihood Estimator (MLE).

Key Insights and Ideas

  • The Power of $1/r$: By representing distance as inverse-range, the "far-field" problem is mathematically neutralized. You no longer need a priori knowledge of whether a source is near or far to choose your model.
  • Efficiency vs. Accuracy: The existing "gold standard" for closed-form MPR (called HGTRS) is accurate but computationally expensive because it solves a complex sub-problem twice. HCO achieves the same accuracy but reduces the computational load by about 14.3%.
  • Robustness to Noise: Interestingly, the authors found that their closed-form solution actually outperforms the Maximum Likelihood Estimator (MLE) in high-noise environments. While MLE is theoretically the "best," it often fails to find the global optimum when the signal is messy; HCO stays on track.

Key Results

  • Accuracy: HCO hits the Cramér-Rao Bound (CRB)—the physical limit of how accurate an estimate can be—under mild noise conditions.
  • Stability: Unlike the HSUM method, which becomes "ill-conditioned" (mathematically unstable) when a source is in a specific direction, HCO remains stable across all orientations.
  • Speed: In CPU time tests, HCO was consistently faster than the high-performing HGTRS and orders of magnitude faster than Semidefinite Relaxation (SDR) methods.
  • Thresholding: The simulations confirmed that while Cartesian-based methods "break" and lose accuracy at certain ranges, the MPR-based HCO maintains a steady performance curve.

Limitations and Future Work

  • Gaussian Noise Assumption: The math assumes "mild Gaussian noise." In real-world urban environments, "multipath" interference (signals bouncing off buildings) creates non-Gaussian noise that might require further algorithm tuning.
  • Static Sources: The paper focuses on a stationary source. The authors suggest that future work could extend this to moving targets by incorporating Frequency Difference of Arrival (FDOA) measurements.
  • Unknown Constants: The current model assumes the speed of signal propagation is known. Extending the MPR framework to handle unknown propagation speeds (common in underwater or complex atmospheric conditions) is another potential research path.

Summary updated

Abstract

Classical localization methods use Cartesian or Polar coordinates, which require a priori range information to determine whether to estimate position or to only find bearings. The modified polar representation (MPR) unifies near-field and far-field models, alleviating the thresholding effect. Current localization methods in MPR based on the angle of arrival (AOA) and time difference of arrival (TDOA) measurements resort to semidefinite relaxation (SDR) and Gauss-Newton iteration, which are computationally complex and face the possible diverge problem. This paper formulates a pseudo linear equation between the measurements and the unknown MPR position, which leads to a closed-form solution for the hybrid TDOA-AOA localization problem, namely hybrid constrained optimization (HCO). HCO attains Cramér-Rao bound (CRB)-level accuracy for mild Gaussian noise. Compared with the existing closed-form solutions for the hybrid TDOA-AOA case, HCO provides comparable performance to the hybrid generalized trust region subproblem (HGTRS) solution and is better than the hybrid successive unconstrained minimization (HSUM) solution in large noise region. Its computational complexity is lower than that of HGTRS. Simulations validate the performance of HCO achieves the CRB that the maximum likelihood estimator (MLE) attains if the noise is small, but the MLE deviates from CRB earlier.
BibTeX
@article{cong2023localization,
  author = {Xunchao Cong and Yimao Sun and Yanbing Yang and Lei Zhang and Liangyin Chen},
  title = {{Localization in Modified Polar Representation: Hybrid Measurements and Closed-Form Solution}},
  journal = {Journal of Systems Engineering and Electronics},
  year = {2023},
  doi = {10.23919/jsee.2023.000146},
}

Loading related content...

Figure