Harris hawks optimization algorithm and BP neural network for ultra-wideband indoor positioning

Area of Science:

• Wireless Communication
• Artificial Intelligence
• Localization Technologies

Background:

• Traditional Back Propagation Neural Networks (BPNN) offer improved accuracy for ultrawideband (UWB) indoor localization but are prone to local minima.
• Optimizing BPNNs is crucial for enhancing UWB indoor positioning performance, especially in non-line-of-sight (NLOS) conditions.

Purpose of the Study:

• To enhance ultrawideband (UWB) indoor positioning accuracy and non-line-of-sight (NLOS) resistance.
• To overcome the limitations of traditional BPNNs by optimizing their weights and thresholds using a metaheuristic algorithm.

Main Methods:

• The study employs the Harris Hawks Optimization (HHO) algorithm to optimize the random weights and thresholds of BPNNs.
• A hybrid HHO-BP algorithm was developed and tested in a two-dimensional localization scenario with four base stations.

Main Results:

• The HHO-BP algorithm demonstrated superior performance compared to the conventional BPNN in both line-of-sight (LOS) and NLOS environments.
• In LOS conditions, HHO-BP reduced the total mean error to 4.50 cm, a 22.57% improvement over conventional BPNN.
• In NLOS conditions, HHO-BP achieved a total mean error of 9.56 cm, a 17.54% improvement over conventional BPNN.

Conclusions:

• The HHO-BP hybrid algorithm significantly enhances UWB indoor localization accuracy and stability.
• This approach offers a viable solution for applications demanding high positional precision, particularly in challenging NLOS environments.