[Paper Review] Localization in Wireless Sensor Grids

Chen Zhang, Ted Herman|ArXiv.org|Jan 25, 2006

Indoor and Outdoor Localization Technologies19 references3 citations

TL;DR

This paper proposes a simple, distributed table-lookup localization algorithm for wireless sensor grids using RSSI-based distance estimation with fuzzy membership functions to improve accuracy despite high ranging errors (up to 60%). It achieves 86% localization accuracy in real-world experiments by leveraging grid topology constraints and minimizing computational overhead, demonstrating robustness in noisy, grassy terrain with minimal hardware requirements.

ABSTRACT

This work reports experiences on using radio ranging to position sensors in a grid topology. The implementation is simple, efficient, and could be practically distributed. The paper describes an implementation and experimental result based on RSSI distance estimation. Novel techniques such as fuzzy membership functions and table lookup are used to obtain more accurate result and simplify the computation. An 86% accuracy is achieved in the experiment in spite of inaccurate RSSI distance estimates with errors up to 60%.

Motivation & Objective

• To investigate whether grid topology simplifies the self-localization problem in wireless sensor networks.
• To develop a lightweight, distributed localization algorithm suitable for resource-constrained sensor nodes.
• To improve localization accuracy using RSSI ranging despite high measurement variance and errors.
• To evaluate the impact of anchor placement and segment size on localization performance.
• To explore error tolerance in 1-hop neighborhood detection and its effect on final localization accuracy.

Proposed method

• Uses grid topology with predefined anchor positions to constrain node localization and reduce search space.
• Employs RSSI-based distance estimation with fuzzy membership functions to classify 1-hop neighbor relationships more robustly.
• Applies table lookup based on hop counts to assign grid coordinates without floating-point arithmetic.
• Introduces a three-stage process: 1) RSSI aggregation, 2) fuzzy membership scoring, 3) 1-hop neighbor classification and grid assignment.
• Uses a grid distance formula: dist((a,b),(c,d)) = max( (|a−c| + |b−d|)/2 , |b−d| ) to compute shortest path hops.
• Implements a distributed architecture where each segment can run localization independently, reducing reliance on central basestations.

Experimental results

Research questions

• RQ1 Can a grid topology enable a simpler, more efficient localization algorithm compared to arbitrary topologies?
• RQ2 How does RSSI-based ranging with high error (up to 60%) affect localization accuracy in real-world conditions?
• RQ3 To what extent can fuzzy membership functions and table lookup reduce the impact of ranging inaccuracies?
• RQ4 What is the trade-off between type A and type B errors in 1-hop neighbor detection, and how does it affect final localization?
• RQ5 How does segment size influence localization accuracy and error accumulation in grid-based localization?

Key findings

• The algorithm achieved 86.0% accuracy in identifying correct 1-hop neighbors in a real-world 50-node grid deployment.
• With 17 type B errors and 9 type A errors, the algorithm maintained high localization fidelity despite significant RSSI ranging variance.
• The use of fuzzy membership functions reduced the number of type A and B errors compared to direct RSSI thresholding, which had only 72.2% correct 1-hop neighbor recognition.
• Larger segment sizes led to decreased accuracy due to cumulative errors in hop counting over longer paths.
• Simulations showed that fewer than 12 type A errors and fewer than 31 type B errors are required for acceptable localization performance.
• The algorithm is robust to high RSSI error (up to 60%) and maintains good performance even with inaccurate distance estimates, thanks to grid constraints and fuzzy logic.