Locally Linear Embedding based Indoor Localization in Internet of Things

Abstract

With the upcoming of smart cities, numerous indoor localization applications plays a significant role. In outdoor, a Global Positioning System (GPS) is majorly used as it’s easy to deploy and provides high accuracy. However, in indoor localization accuracy becomes a challenge due to poor signal strength. This invokes the necessity for a mechanism to get precise node location. In this paper, a multilateration and Locally Linear Embedding (LLE) based localization approach is proposed. In the proposed mechanism, distance-RSSI characterization is done at the initial stage to obtain distances between each pair of sensor nodes. The multilateration method is used to obtain the course grain location of sensor nodes. Finally, LLE is applied to refine locations. Simulation results show that the proposed mechanism is robust and accurately localizes sensor nodes as compared to existing algorithms in an indoor environment.