A Scalable Localization System for Critical Controlled Wireless Sensor Networks



Determining the positions of unknown position nodes, especially mobile nodes in a wireless sensor network (WSN), is critical for many applications. It helps to identify the location of the collected data and of the node carrier such as a worker, patient or vehicle. This information is often critical on supporting the right (time) decisions. This paper presents a scalable localization system targeting Controlled WSNs for critical industrial environments. Multiple positioning methods were implemented and evaluated using real testbeds setting up in both laboratory and industrial environments. The measurement used in our localization system is Received Signal Strength Indicator (RSSI). Although it is unstable and with high variance, the experimental results show that pattern matching based methods such as k-nearest neighbors, probability-based (Bayesian Theorem) and Kalman filter over probability-based produce an acceptable accuracy that is sufficient for many applications. In particular, the average distance error of 3.37m can be achieved with 50th and 80th percentile distance errors of 2 and 5.35m respectively. In addition, by carefully designing the positions of beacons it is possible to obtain the average distance error about 2.23m and 50th and 80th percentile distance errors of 0.46 and less than 4.4m respectively.


Localization, Wireless Sensor Network, Baysian method, K-Nearest Neighbors


Localiation in Wireless Sesnsor Networks

Related Project

iCIS - Intelligent Computing in the Internet of Services


ICUMT 2014 - the 6th International Congress on Ultra Modern Telecommunications and Control Systems, October 2014

PDF File

Cited by

No citations found