Location Optimization of Wireless Sensor Network in Intelligent Workshop Based on the Three-Dimensional Adaptive Fruit Fly Optimization Algorithm

Shaobo Li, Chenglong Zhang, Jinglei QuInt. J. Online eng 2018 · DOI

Introduction

In the rapidly evolving landscape of the "Internet of Manufacturing Things," real-time asset tracking—encompassing staff, materials, and equipment—is a cornerstone of the intelligent workshop. However, traditional localization methods like GPS fail in indoor industrial environments due to signal blockage and interference. While Wireless Sensor Networks (WSNs) offer a solution, standard localization algorithms often struggle with "local optima," where they settle on a suboptimal coordinate because it appears better than its immediate surroundings. Furthermore, most existing research is confined to 2D planes, failing to account for the vertical dimension critical in modern high-density manufacturing environments.

The 3D-AFOA Methodology

To address these challenges, this research proposes the Three-Dimensional Adaptive Fruit Fly Optimization Algorithm (3D-AFOA). This approach integrates WSNs into a broader enterprise architecture, connecting physical manufacturing resources with high-level management systems like ERP and MES.

System architecture of the intelligent workshop integrating WSNs with enterprise management systems.

The algorithm mimics the foraging behavior of fruit flies, which use a sense of smell for global searching and vision for precise local positioning. The "Adaptive" component is derived from Particle Swarm Optimization (PSO), introducing an adaptive inertial weight coefficient ($\omega$).

In the early stages of the search, the weight is high, encouraging the algorithm to explore the entire 3D space. As the process continues, the weight decreases, allowing the algorithm to focus on fine-tuning the coordinates. This dynamic adjustment prevents the algorithm from getting stuck in local optima and ensures a more accurate global minimum for location error.

Experimental Results

The performance of the 3D-AFOA was evaluated using a MATLAB simulation of a virtual workshop measuring 80m x 30m x 6m. The study compared the proposed adaptive algorithm against the standard 3D Fruit Fly Optimization Algorithm (3D-FOA).

Comparison of location error convergence between the standard 3D-FOA and the proposed 3D-AFOA.

As shown in the convergence analysis, the 3D-AFOA consistently achieves lower error rates and faster convergence than the standard version. Specifically, with 3 anchor nodes, the 3D-AFOA reached a minimum location error of 1.54 cm in 179 iterations, whereas the standard 3D-FOA required 475 iterations to reach a less accurate result of 1.73 cm.

The research also quantified the impact of increasing the number of anchor nodes (fixed sensors). The results are summarized below:

Number of Anchor Nodes Minimum Location Error Generation of Convergence
3 1.54 cm 179
4 1.38 cm 73
5 1.13 cm 44

Increasing the sensor density from 3 to 5 nodes not only improved accuracy to approximately 1.1 cm but also slashed the computational time required for convergence by over 75%.

Conclusion and Future Work

The 3D-AFOA demonstrates a significant advancement in indoor localization for smart factories, providing centimeter-level accuracy in a 3D environment while maintaining high computational efficiency. By successfully navigating the local optima problem, it provides a robust foundation for real-time asset management.

Future research will focus on transitioning from simulations to real-world physical environments to account for electromagnetic interference and physical obstacles. Additionally, the authors plan to investigate the algorithm's performance in high-density node environments and its feasibility on low-power hardware typical of industrial WSN nodes.

Summary updated

BibTeX
@article{li2018location,
  author = {Shaobo Li and Chenglong Zhang and Jinglei Qu},
  title = {{Location Optimization of Wireless Sensor Network in Intelligent Workshop Based on the Three-Dimensional Adaptive Fruit Fly Optimization Algorithm}},
  journal = {Int. J. Online eng},
  year = {2018},
  doi = {10.3991/ijoe.v14i11.9544},
}
Figure