Abstract:Location fingerprinting based on RSSI becomes a mainstream indoor localization technique due to its advantage of not requiring the installation of new infrastructure and the modification of existing devices, especially given the prevalence of Wi-Fi-enabled devices and the ubiquitous Wi-Fi access in modern buildings. The use of AI/ML technologies like DNNs makes location fingerprinting more accurate and reliable, especially for large-scale multi-building and multi-floor indoor localization. The application of DNNs for indoor localization, however, depends on a large amount of preprocessed and deliberately-labeled data for their training. Considering the difficulty of the data collection in an indoor environment, especially under the current epidemic situation of COVID-19, we investigate three different methods of RSSI data augmentation based on Multi-Output Gaussian Process (MOGP), i.e., by a single floor, by neighboring floors, and by a single building; unlike Single-Output Gaussian Process (SOGP), MOGP can take into account the correlation among RSSI observations from multiple Access Points (APs) deployed closely to each other (e.g., APs on the same floor of a building) by collectively handling them. The feasibility of the MOGP-based RSSI data augmentation is demonstrated through experiments based on the state-of-the-art RNN indoor localization model and the UJIIndoorLoc, i.e., the most popular publicly-available multi-building and multi-floor indoor localization database, where the RNN model trained with the UJIIndoorLoc database augmented by using the whole RSSI data of a building in fitting an MOGP model (i.e., by a single building) outperforms the other two augmentation methods as well as the RNN model trained with the original UJIIndoorLoc database, resulting in the mean three-dimensional positioning error of 8.42 m.

Analyzing Machine Learning Models with Gaussian Process for the Indoor Positioning System

Localization Reliability Improvement Using Deep Gaussian Process Regression Model

GP-SLAM plus : real-time 3D lidar SLAM based on improved regionalized Gaussian process map reconstruction

Evaluation of GP-SLAM in Real-World Environments

An Indoor Wi-Fi Localization Algorithm Using Ranging Model Constructed With Transformed RSSI and BP Neural Network

GM(1,1)-Based Weighted K-Nearest Neighbor Algorithm for Indoor Localization

Multi-Dimensional Wi-Fi Received Signal Strength Indicator Data Augmentation Based on Multi-Output Gaussian Process for Large-Scale Indoor Localization

SGP-RI: A Real-Time-Trainable and Decentralized IoT Indoor Localization Model Based on Sparse Gaussian Process with Reduced-Dimensional Inputs

Gaussian Processes for Analyzing Positioned Trajectories in Sports

Learning While Navigating: A Practical System Based on Variational Gaussian Process State-Space Model and Smartphone Sensory Data

Improving smartphone GNSS positioning in challenging urban environments using GA-BPNN

Augmentation of Fingerprints for Indoor WiFi Localization Based on Gaussian Process Regression

Precise WiFi Indoor Positioning using Deep Learning Algorithms

Enhanced Indoor Positioning Using RSSI and Time-Distributed Auto Encoder-Gated Recurrent Unit Model

Multi-Output Gaussian Process-Based Data Augmentation for Multi-Building and Multi-Floor Indoor Localization

Research on Indoor Position Fingerprint Location Based on Machine Learning Combined Particle Filter

Improving Indoor WiFi Localization by Using Machine Learning Techniques

An Improved BPNN Method Based on Probability Density for Indoor Location

Uncertainty in Position Estimation Using Machine Learning

Using the MNL Model in a Mobile Device's Indoor Positioning

An RFID Indoor Positioning Algorithm Based on Bayesian Probability and K-Nearest Neighbor