Improving the Performance of RSSI Based Indoor Localization Techniques Using Neural Networks

Node localization is an essential part of Wireless sensor network and has a good scope for research and development. Many revolutionary ideas like driverless cars, augmented reality and instant emergency response systems are dependent on precise localization. Localization in an indoor environment is not generic and simple as in outdoors due to the increased randomness, attenuation, heterogeneity and interference. These factors reduce the precision of popular localization algorithms in an indoor environment. This paper discusses about error reduction in a RSSI based localization algorithm using neural networks. Parallel computational capabilities and non-linearity of neural networks would come in handy with the constraints in indoor localization. In-depth discussion has been made in this paper about the procedure followed for localization, sources of error and error controlling mechanisms applied. Simulation results are also discussed towards the end, which show significant improvement in localization performance with the error correction mechanism.