A New RSSI-based Centroid Localization Algorithm by Use of Virtual Reference Tags

Joan Lu, Hua Long, Qiang Xu, Qian LeiUniversity 2013 · OpenAlex

Abstract: A good design of node location is critical for efficient and effective wireless communications. This paper presents an improved algorithm, in order to solve the low localization accuracy caused by traditional centroid algorithm. The improved algorithm combined with VIRE system and traditional centroid algorithm. The VIRE algorithm is introduced and the signal propagation model is utilized to construct virtual reference tags in the location area. Simulation shows that this further developed algorithm has further improved the accuracy of positioning up to 35.12% compared to the traditional centroid algorithm. It is concluded that this algorithm can further improve the locating accuracy in comparison with the original centroid algorithm.

Introduction

In Wireless Sensor Networks (WSNs), knowing the exact location of a node is fundamental for data routing and environmental monitoring. However, localization is a balancing act between cost and precision. Traditional "centroid" algorithms are popular because they are computationally light and don't require expensive hardware, but they are often inaccurate. Conversely, high-precision methods typically require a dense deployment of physical reference tags or specialized hardware for measuring Time of Arrival (ToA).

This paper introduces an improved localization algorithm that significantly boosts positioning accuracy without requiring additional hardware. By combining the Virtual Reference Elimination (VIRE) system with a "renewed" centroid calculation that narrows down the search area into sub-regions, the proposed method achieves a 35.12% improvement in accuracy over traditional centroid methods.

Methodology

The proposed approach leverages a two-stage process using virtual infrastructure to enhance resolution without increasing physical costs.

Virtual Reference Elimination (VIRE)

Instead of deploying hundreds of physical RFID tags, the system uses a signal propagation model to create a dense grid of "virtual reference tags." By calculating theoretical Received Signal Strength Indicator (RSSI) values at these grid points, the system simulates a high-density environment. When an unknown node needs to be located, the system identifies the four virtual tags whose theoretical RSSI values most closely match the node's actual measured RSSI.

The grid layout of physical reference tags (white circles) and the generated virtual reference tags (grey circles).

The Renewed Centroid Algorithm

The core innovation lies in the "renewed" centroid calculation. In a standard centroid algorithm, one would simply average the coordinates of the four nearest tags. This improved version divides the square area formed by those four tags into smaller sub-regions (labeled U, M, V, and W) and segments.

The division of the localization cell into sub-regions (M, V, U, W) to refine the estimated position.

By comparing the relative differences (deltas) between the node's RSSI and the tags' RSSI, the algorithm intelligently selects a more precise center point within a specific sub-region rather than just the center of the whole square. This effectively doubles or triples the resolution of the grid.

Experimental Results

The study utilized simulated data generated in Matlab 7.6.0 within a 90x90 unit area with four readers placed at the corners. Performance was benchmarked against the Traditional Centroid Algorithm across ten specific object nodes.

Key findings include:

  • Accuracy Boost: The renewed algorithm improved positioning accuracy by up to 35.12%.
  • Optimal Grid Configuration: The researchers discovered a "sweet spot" for density. Increasing the number of virtual tags doesn't always lead to better results; they found an optimal point (49 virtual tags in their 90x90 setup) where accuracy is maximized before redundant calculations yield diminishing returns.
  • Consistency: In simulations of 100 tests, the renewed algorithm consistently produced lower error margins across all tested object nodes.

Discussion and Conclusion

The most powerful insight from this research is that a well-tuned signal propagation model can act as a substitute for hardware density. This allows for high-resolution mapping without the logistical burden of installing thousands of physical sensors.

However, the authors acknowledge limitations. The current model assumes a stable signal propagation environment. In real-world indoor settings, "multipath effects" and physical obstacles can distort RSSI values. Future work will focus on refining signal models to handle complex, cluttered environments and optimizing the selection logic for virtual tags to prevent compounded errors from initial signal noise.

Summary updated

Abstract

A good design of node location is critical for efficient 
\nand effective wireless communications. This paper presents an 
\nimproved algorithm, in order to solve the low localization 
\naccuracy caused by traditional centroid algorithm. The 
\nimproved algorithm combined with VIRE system and 
\ntraditional centroid algorithm. The VIRE algorithm is 
\nintroduced and the signal propagation model is utilized to 
\nconstruct virtual reference tags in the location area. Simulation shows that this further developed algorithm has further improved the accuracy of positioning up to 35.12% compared 
\nto the traditional centroid algorithm. It is concluded that this algorithm can further improve the locating accuracy in comparison with the original centroid algorithm.
BibTeX
@inproceedings{lu2013new,
  author = {Joan Lu and Hua Long and Qiang Xu and Qian Lei},
  title = {{A New RSSI-based Centroid Localization Algorithm by Use of Virtual Reference Tags}},
  booktitle = {University},
  year = {2013},
}
Figure