Research on the Construction of Radio-Map Based on Support Vector Regression

Abstract

Location fingerprinting in WLAN has become the mainstream of indoor positioning research in recent years. The Radio-map constructed during offline phase is an important basis for positioning, but it costs huge human and computational resources. In order to improve efficiency and reduce the labor cost of establishing the Radio-map during offline phase, this paper proposes a method based on support vector regression (SVR) algorithm to build a Radio-map. In this method, a nonlinear mapping between RSS and physical locations is constructed by SVR. Experimental results show that, this method can reduce the workload of Radio-map construction and ensure positioning accuracy at the same time.