A Parallel Artificial Neural Network Learning Scheme Based on Radio Wave Fingerprint for Indoor Localization
- Resource Type
- Conference
- Authors
- Park, Chan Uk; Shin, Hong-Gi; Choi, Yong-Hoon
- Source
- 2018 Tenth International Conference on Ubiquitous and Future Networks (ICUFN) Ubiquitous and Future Networks (ICUFN), 2018 Tenth International Conference on. :794-797 Jul, 2018
- Subject
- Communication, Networking and Broadcast Technologies
Computing and Processing
Signal Processing and Analysis
Fingerprint recognition
Neural networks
Machine learning
Euclidean distance
Two dimensional displays
Three-dimensional displays
Floors
Indoor Positioning
Aritificial Neural Network (ANN)
Deep Learning
Data Augmentation
- Language
- ISSN
- 2165-8536
Radio wave fingerprinting is known to be the best method for indoor positioning, and its performance depends greatly on the data comparison algorithm that is used. This paper implements a radio wave fingerprint positioning method with artificial neural network learning to improve the performance of a conventional radio fingerprint positioning algorithm based on the Euclidean distance. We propose a parallel learning method to reduce the error in the indoor height and an indoor positioning data augmentation method for data generalization. This method exhibits a higher performance than an existing Euclidean distance based positioning method. In particular, the data augmentation technique can be applied without depending on the specific positioning algorithm.