Abstract:Among various ranging techniques, Radio Signal Strength (RSS) based approaches attract intensive research interests because of its low cost and wide applicability. RSS-based ranging is prone to be affected by the multipath phenomenon which allows the radio signals to reach the destination through multiple propagation paths. To address this issue, previous works try to profile the environment and refer this profile during run-time. In a practical dynamic environment, however, the profile frequently changes and the painful retraining is needed. Rather than such static ways of profiling the environments, in this paper, we try to accommodate the environmental dynamics automatically in real-time. The key observation is that given a pair of nodes, the RSS at different spectrum channels will be different. This difference carries the valuable phase information of the radio signals. By analyzing these RSS values, we are able to identify the amplitude of signals solely from the Line-of-Sight (LOS) path. This LOS amplitude is a simple function of the path length (the physical distance). We find that the analysis is a typical non-linear curvature fitting problem that has no general routing algorithms. We prove that, this problem format is ill-conditioned which has no stable and trustable solutions. To deal with this issue, we further explore the practical considerations for the problem and modify it to a greatly improved conditioning shape. We solve the problem by numerical iterations and implement these ideas in a real-time indoor tracking system called MuD. MuD employs only three TelosB nodes as anchors. The experiment results show that in a dynamic environment where five people move around, the averaged localization error is about 1 meter. Compared with the traditional RSS-based approaches in dynamic environments, the accuracy improves up to 10 times.

RSSI Distance Model Based on BP Neural Network

Multi-parameter Online Measurement IoT System Based on BP Neural Network Algorithm.

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

A Machine Learning based Intelligent Propagation Model for RSRP prediction

Research on Indoor Positioning Algorithm Based on SAGA-BP Neural Network

A new deep learning-based distance and position estimation model for range-based indoor localization systems

An Improved BPNN Method Based on Probability Density for Indoor Location

Base on Wireless Sensor Network Improve RSSI Mine Location Algorithm Design and Realize

Using Low-Power Radios for Mobile Robots Navigation

RSSI-BASED ADAPTIVE WIRELESS POSITIONING ALGORITHM

Weighted Centroid Localization Algorithm Based on MEA-BP Neural Network and DBSCAN Clustering

Experimental Study of Bluetooth Indoor Positioning Using RSS and Deep Learning Algorithms

A Model-Based BLE Indoor Positioning System Using Particle Swarm Optimization

Position measurement using Bluetooth

Distance Measurement Model Based on RSSI in WSN.

Research on indoor and outdoor positioning switching algorithm based on improved PSO-BP

Probabilistic Neural Network for RSS-Based Collaborative Localization.

Research Progress of Wireless Positioning Methods Based on RSSI

Rss-Based Ranging by Leveraging Frequency Diversity to Distinguish the Multiple Radio Paths

Human Detection through RSSI Processing with Packet Dropout in Wireless Sensor Network

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