Optimising mobile phone self-location estimates by introducing beacon characteristics to the algorithm

Positioning technologies that use Global System for Mobile Communication (GSM) networks for location estimation (such as the Privacy Observant Location System (POLS) and the Place Lab framework) lack the accuracy that other positioning technologies like the Global Positioning System (GPS) have. GPS receivers are capable most of the time of placing a person within 10 metres of a known location. Place Lab is an open platform framework implemented in Java for client-side location sensing that can calculate a position estimate from various beacon sources, such as GSM beacons. The POLS framework is a counterpart of Place Lab for Windows Smartphone devices which provide the tools to develop location-based services quickly. There is a lack of accuracy however when the location estimation algorithm uses only GSM readings. Measurements that have been made with Place Lab show a median accuracy of 232m in downtown areas. Place Lab and POLS do not need additional hardware, apart from the mobile phone itself however their lack of accuracy compared to GPS is significant. Due to this rather poor accuracy, the use of those frameworks is limited to applications where the accuracy is not crucial. This paper presents the results of improving the accuracy of location estimation in urban areas by extending the algorithm used in the POLS and Place Lab frameworks to take into account the beacon properties, Effective Radiated Power (ERP) and beacon height when estimating a position. The extended algorithm based on beacon properties outperforms the centroid algorithm by over 30%.