Wou Onn Choo1, Lam Hong Lee2, Yen Pei Tay2, Khang Wen Goh2, Wen Yeen Chue3, Suliman Mohamed Fati1


1Faulty of Information Technology and Science, INTI International University, Nilai, Negeri Sembilan, 71800, Malaysia

2Faculty of Science and Technology, Quest International University Perak, 30250 Ipoh, Perak, Malaysia.

3Faculty of Business and Information Science, UCSI University, 56000 Kuala Lumpur, Malaysia.


Corresponding Author:,


An intelligent agent equipped with cellular-assisted Global Positioning System (GPS) positioning has been proposed in this paper. The positioning technique has been enhanced by using Bayesian and Self-Organizing Maps approaches. Due to the overlapping of coverage areas of cellular towers, conventional cellular-based positioning techniques have been reported to be inaccurate. Current cellular-assisted GPS positioning techniques are cost ineffective as extensive investments on hardware deployments are required in order to achieve the highly accurate positioning performance. A relatively low cost approach is presented in this paper for more economical and satisfactory cellular-assisted GPS positioning. Raw location information, in the form of cellular identity (ID) and GPS coordinate pairs, are acquired by using equipment such as smart phones and GPS trackers. These raw information were categorized into categories according to the distribution patterns of cellular towers. The cellular ID and GPS coordinate pairs were further grouped within each of the individual cellular IDs. An intelligent software agent equipped with data mining capabilities was then deployed to collect and process the device-coordinates in order to predict the optimal GPS coordinates of the cellular ID. This results to the determination of a virtual cellular tower location for each cellular ID, to provide more precise location positioning. Experimental results show that the prediction of location using GPS coordinate of cellular IDs helps in improving the contemporary cellular-assisted GPS positioning technique to sub-kilometre accuracy.

Keywords: Cellular-Assisted GPS Positioning; Bayesian; Self-Organizing Map