Abstract
Errors are caused by wall reflection and noise in the visible light indoor positioning system. Some indoor positioning algorithms based on fingerprint are compared through simulation. An improved indoor positioning algorithm that combines K-Nearest Neighbor and Bayesian theory algorithm is proposed. Firstly, several grid points which are close to the receiver's position are selected by KNN algorithm, then the posterior probability is calculated by Bayesian algorithm. The point with the largest posterior probability is the estimated position. The improved algorithm simplifies the Bayesian algorithm and improves the positioning accuracy of KNN algorithm, with an average error of 0.27m.