Fuzzy System and Improved APIT (FIAPIT) Combined Range-free Localization Method for WSN

Xiaofeng Li, Liangfeng Chen, Jianping Wang, Zhong Chu, Qiyue Li, Wei SunKSII 2015 · DOI

Abstract

Among numerous localization schemes proposed specifically for Wireless Sensor Network (WSN), the range-free localization algorithms based on the received signal strength indication (RSSI) have attracted considerable research interest for their simplicity and low cost. As a typical range-free algorithm, Approximate Point In Triangulation test (APIT) suffers from significant estimation errors due to its theoretical defects and RSSI inaccuracy. This paper introduces FIAPIT, a hybrid method that integrates an improved geometric test with a fuzzy logic system to enhance localization accuracy in noisy environments.

Introduction: The Challenge of WSN Localization

In a Wireless Sensor Network, knowing the precise location of a sensor node is often as critical as the environmental data it collects. While GPS provides high accuracy, its cost and power consumption are prohibitive for large-scale deployments of tiny, low-cost sensors. Consequently, researchers favor "range-free" localization methods. These algorithms estimate positions based on connectivity and signal strength relative to "anchor" nodes—nodes with pre-determined coordinates.

A prominent range-free method is the Approximate Point In Triangulation (APIT) test. APIT determines if a sensor node is located within various triangles formed by sets of three anchor nodes. By intersecting these triangles, the algorithm narrows down the possible location of the sensor. However, APIT is plagued by two primary issues: theoretical geometric defects and the inherent unreliability of Received Signal Strength Indication (RSSI).

The Theoretical Defects of APIT

The standard APIT algorithm relies on a simple geometric test: if a node can move in any direction and find a position where it is further from (or closer to) all three anchors of a triangle simultaneously, it is considered outside that triangle. In practice, this test frequently fails due to the discrete nature of node placement.

InToOut and OutToIn errors in the standard APIT algorithm.

As shown in the figure above, APIT suffers from two main types of errors:

  1. InToOut Error: A node is actually inside the triangle, but due to the limited number of neighbor nodes available for comparison, the algorithm incorrectly concludes it is outside.
  2. OutToIn Error: A node is outside the triangle, but the test fails to find a direction that satisfies the "further/closer" condition, leading to an incorrect "inside" classification.

Addressing RSSI Inaccuracy with Fuzzy Logic

Beyond geometric flaws, APIT's reliance on RSSI is a major source of error. In ideal conditions, signal strength decreases predictably with distance. In reality, environmental factors like multipath fading, reflection, and obstacles create a "noisy" relationship where signal strength is an unreliable proxy for distance.

RSSI vs. Distance relationship showing significant signal noise.

The empirical data collected by the researchers (shown above) illustrates that at any given distance, the RSSI can vary significantly. Standard APIT uses a binary "inside or outside" decision based on these noisy values, which often leads to catastrophic localization failures when a single reading is slightly off.

The FIAPIT Solution

To overcome these hurdles, the authors propose FIAPIT (Fuzzy system and Improved APIT). This method improves localization through two main innovations:

1. Improved APIT (IAPIT)

The researchers refined the underlying geometric logic by introducing a "Point out of Triangulation" (POT) test for nodes near the triangle edges and a distance threshold for "far" cases. This allows the algorithm to better distinguish whether a neighbor node is truly further from or closer to the anchors, specifically targeting the reduction of "InToOut" errors in dense networks.

2. Fuzzy Inference Engine

Instead of a rigid binary output, FIAPIT employs a fuzzy logic system. It maps the noisy RSSI differences into membership degrees (such as "Far," "Near," or "Static"). This allows the system to handle uncertainty gracefully. Rather than discarding a triangle because a single noisy reading suggests the node is "outside," the fuzzy system assigns a weight to that triangle's contribution.

The final position is calculated as a weighted average of the centers of gravity (COGs) of the various triangle intersection areas. The weights are dynamically determined by the fuzzy system, ensuring that the algorithm "trusts" more reliable geometric overlaps.

Key Results and Insights

The performance of FIAPIT was evaluated using Network Simulator 2 (NS2) and calibrated with 6,000 real-world RSSI measurements. The results highlight several key advantages:

  • Robustness to Signal Irregularity: FIAPIT significantly outperformed both the original APIT and Weighted Centroid Localization (WCL) in scenarios with a high Degree of Irregularity (DOI).
  • Density Performance: The IAPIT component proved particularly effective in high-density networks, where it successfully corrected the geometric errors that typically degrade standard APIT performance.
  • Stability: FIAPIT demonstrated greater stability across varying Anchor-to-Node Range (ANR) ratios, maintaining accuracy even when anchors were sparse.

Conclusion and Future Work

FIAPIT represents a significant step forward in range-free localization by acknowledging that binary decisions are the enemy of robustness in noisy environments. By combining geometric refinements with fuzzy logic, the system provides a more accurate and stable location estimate. Future research may explore using machine learning to further optimize weights or developing more energy-efficient communication protocols for the necessary beacon exchanges.

Summary updated

Abstract

Among numerous localization schemes proposed specifically for Wireless Sensor Network (WSN), the range-free localization algorithms based on the received signal strength indication (RSSI) have attracted considerable research interest for their simplicity and low cost. As a typical range-free algorithm, Approximate Point In Triangulation test (APIT) suffers from significant estimation errors due to its theoretical defects and RSSI inaccuracy.
BibTeX
@inproceedings{li2015fuzzy,
  author = {Xiaofeng Li and Liangfeng Chen and Jianping Wang and Zhong Chu and Qiyue Li and Wei Sun},
  title = {{Fuzzy System and Improved APIT (FIAPIT) Combined Range-free Localization Method for WSN}},
  booktitle = {KSII},
  year = {2015},
  doi = {10.3837/tiis.2015.07.005},
}
Figure