Indoor location research based on AP layout optimization and K-means clustering algorithm

Abstract

The traditional clustering algorithm passively depends on the number of Access Points(AP) deployed on indoor positioning environment，which leads to low efficiency and high positioning error. The layout of AP is a key factor which affects the positioning accuracy of indoor location fingerprint positioning. So a sensor network is built in this paper, which consists of the Intel chips embedded micro-system and the SA44B measuring receivers produced by Signal Hound US. Firstly, the objective function of indoor positioning is established on the basis of the wave path loss theory. Next, the simulated annealing algorithm and the simplex fusion algorithm are used to optimize the objective function, and then the most reasonable layout of AP indoor location is achieved. Finally, the optimized AP position coordinates as the initial cluster centers that are modified by the K-means clustering algorithm，to improve the positioning efficiency and the precision of the system. The traditional K-means algorithm is used as the comparison object in the paper. The experimental results show that the precision of the clustering localization algorithm after the AP location optimization is improved by 13.8%.

Keywords

• indoor location
• AP position optimization
• simulated annealing algorithm
• simplex method
• embedded system
• spectrum analyzer and measuring receiver