Accurate visible light positioning technique using extreme learning machine and meta-heuristic algorithm

Since input weights and hidden biases affect the overall performance of extreme learning machine (ELM) based on a single-hidden layer, two meta-heuristic algorithms of grey wolf optimizer (GWO) and particle swarm optimization (PSO) are adopted to optimize the input weights and hidden biases of ELM respectively in order to obtain higher classification accuracy. Namely, ELM-GWO and ELM-PSO are two enhanced ELM algorithms. Compared with the other existing techniques such as the traditional ELM and adaptive boosting (AdaBoost), the positioning performance of two enhanced ELM algorithms is analyzed by simulation and experiment in visible light positioning (VLP) systems. The experimental results show that the probabilities of two-dimensional (2D) positioning error being less than 10 cm for ELM, ELM-GWO, ELM-PSO, and AdaBoost are 84.25%, 90.25%, 93.25%, and 19.75%, respectively. And the probabilities of three-dimensional (3D) positioning error being less than 10 cm for ELM, ELM-GWO, ELM-PSO, and AdaBoost are 45.17%, 84.67%, 80.33%, and 16%, respectively. Both simulation and experimental results show that two enhanced ELM algorithms have better positioning performance and robustness. In addition, although increasing the number of iterations and the number of hidden neurons can effectively reduce the positioning error, the computational complexity is relatively high in large-scale fingerprint positioning scenarios. Therefore, a hybrid positioning method combining enhanced ELM and region division is further proposed. Both simulation and experimental results show that the proposed hybrid method can significantly improve the positioning accuracy while reducing the computational complexity.