Evaluation of Wifi Technologies for Indoor Positioning Applications

For the majority of applications that use positioning information, solutions from Global Navigation Satellite Systems (GNSS) can easily meet the accuracy, availability and reliability required. This assumes that the GNSS receiver is always operating in an environment that offers clear sky visibility. However, for a growing number of location based services and system, operation in urban or indoor environments is essential. Alternative positioning technologies such as WiFi, are gaining popularity due to the ever increasing availability of access points installed for computer networking that can be used simultaneously as an infrastructure for positioning at no additional cost. This paper investigates the potential of WiFi positioning to provide solutions in environments that currently challenge GNSS performance. It presents the practical results generated from a case study undertaken to benchmark commercially available WiFi positioning systems. Details of the tests conducted, and a summary of WiFi positioning performance will be presented in this paper INTRODUCTION The Global Positioning System (GPS) has traditionally underpinned the development of ubiquitous positioning solutions. This stems from the increasing recognition across society that under ideal operational conditions, GPS can meet the majority of attributes of ubiquitous positioning system ie accuracy, reliability and availability. However, it is also widely acknowledged that in certain environments these performance attributes can quickly deteriorate to unacceptable levels, with often only weak GPS signals available. Alternative positioning technologies that can be used in difficult environments include assisted GPS (AGPS) (Bryant, 2004) which exploits weak GPS signals, ultra wide band (UWB) (Fontana et al 2002), wireless local area networks (WiFi) (Hatami & Pavlan, 2006) and Radio Frequency Identification (RFID) (Wang et al, 2007). However, AGPS does not meet the accuracy requirements for many applications eg personnel tracking due to extreme multipathing and signal attenuation effects, and UWB and RFID techniques require significant infrastructure and often costly customisations for A. Kealy, B. Li, T. Gallagher, A. Dempster individual applications. WiFi signals are different in this regard. Over the past five years, tens of millions of WiFi access points have been deployed within networks by individuals, businesses, academic institutions, retail stores, public buildings etc. These networks have been established primarily to facilitate more efficient and flexible access to data and communications. However, these access points repeatedly broadcast a signal the strength of which can be measured and used in algorithms to locate a WiFi enabled device. The increasing pervasiveness of WiFi signals has resulted in the development of several commercial WiFi positioning systems; examples are “Ekahau Positioning Engine (EPE)” and “Skyhook Wireless Positioning System (WPS)”. Unfortunately, the positioning capabilities of WiFi are weakened by the fact that positioning performance is highly dependent on the strength of the signal received from nearby access points, which are affected by a diversity of uncontrollable environmental effects eg people, building material etc. Despite this significant limitation WiFi signals continue to be an attractive alternative positioning technology due to the growing number of WiFi enabled mobile devices being produced (over 360 million WiFi handsets projected to be sold annually by 2011) combined with the rapid growth in WiFi access points being installed globally (shipments of consumer oriented WiFi access points are expected to grow from 6million in 2008 to 88 million in 2013) (ABI Research, 2008). This paper presents the results of a practical investigation into the performance of commercially available WiFi positioning systems and mobile devices. The results depict the typical positioning performance characteristics of WiFi signals and indicate the potential of WiFi to complement GNSS performance within a ubiquitous positioning system without the need for additional costly or customised infrastructure. BACKGROUND GNSS performance in difficult environments such as urban canyons or indoors is characterised by significant multipathing, poor geometry, weak signals or signal unavailability. Whilst high sensitivity GNSS receivers have been developed to use the weak signals that may still be available in these environments, the problem of multipathing and poor satellite geometry can still affect the accuracy and reliability of the position solution. Figure 1 shows a plot of position solutions computed from two commercially available high sensitivity single frequency GPS receivers, the SiRFstarIII (red dots) and ublox Antaris4 (green dots). Figure 1 also shows solutions computed from a dual frequency Leica 1200 GNSS receiver (blue dots). This solution was post processed to generate ambiguity fixed solutions where possible. All data was collected simultaneously on 1 May 2008 at 10:30am for a time period of around 30mins and covering a distance of approximately 5km around the central business district (CBD) in Melbourne. The receivers were mounted on a vehicle and the epoch interval for all receivers was 1 second. Figure 1 also shows the solutions obtained along an enlarged region on Collins Street, the primary business street in the Melbourne CBD comprising multi storey office buildings and car parks. It can be seen that the Leica GNSS receiver was unable to provide precise ambiguity fixed solutions for the majority of time on this street. Even A. Kealy, B. Li, T. Gallagher, A. Dempster with a combined GPS/GLONASS constellation, outages of up to 2mins were experienced with only 27% of the time the receiver able to compute an ambiguity fixed solution. This situation can be explained by looking at Figure 2 which shows the satellite availability for all three receivers during the data collection period. Fig. 1: GNSS positioning performance in a typical difficult environment