HIWL: An Unsupervised Learning Algorithm for Indoor Wireless Localization

An advanced unsupervised learning algorithm for a precise measurement of the local position of an indoor mobile target is proposed. In this work, the indoor wireless localization is addressed with HIWL, an unsupervised learning algorithm based on HMM. The locations of reference nodes and site survey are no longer needed in this algorithm. A sample data process with K-means which helps us produce discrete observation sequences is introduced. Also, a family of equations to compute effective initial parameters of HMM is presented. Experiments show that HIWL can achieve better localization accuracy.